BACKGROUND OF THE INVENTION

Field of the Invention Monoclonal and polyclonal antibodies offer many opportunities for distinguishing between different antigens or antigenic sites. In many cases, even though the anti¬ bodies may have specific binding sites, because of the large number of determinants present in the system of interest, the potential for a variety of amino acid sequences to assume similar polar and spatial confirmations, and various degrees of non-specific binding, even injected antisera specific for a particular epitope may be widely distributed in a host. Particularly, when one is dealing with a system as large and complex as a mammalian system, there is a substantial challenge in developing antibodies which will •provide a high level of discrimination between the site of interest and the large number of other sites which are present in the host. An area of particular interest is detection of breast cancer and its metastases. For appropriate therapy for breast cancer, there must be reliable detection and tumor staging. The ability to detect hitherto unknown metastases and more accurately determine the extent of lymph node involvement can radically change the therapeutic approach. The present techniques are of limited applica¬ tion, permitting detection of only certain metastases based on limitations imposed by size, depth and vascularity. The present techniques have the further failure of being non- specific in not being able to distinguish between the source of the tumor as well as between a tumor and a cyst, abcess or inflammatory site.

Cell-type-specific antisera against mammary epithelial cells which bind to both normal and neoplastic mammary epithelial cells have been reported and shown to localize mouse mammary epithelial carcinomas in mice when labelled with radioactive iodine and injected intravenously

into mice carrying subcutaneously grafted mouse mammary tumors that simulated metastases. Whole-body radionuclide imaging of these mice allowed visual localization of the tumors. However, the prolonged clearance time of the tagged antiserum from the plasma and the profuse vascularization of both breast and non-breast grafted tumors made specific localization of the breast tumors a rather uncertain procedure. Description of the Prior Art Ceriani et al., PNAS USA (1977) 74:582-586 describes cell-type-speci ic antiserum against mammary epithelial cells which bind exclusively to these cells, both normal and neoplastic. See also Peterson et al., Int. J. Cancer (1978) 22:655-661. Ceriani and Peterson, Cell Differentiation (1978) 7:355-366 describe the characteriza¬ tion of mouse mammary epithelial antigens carried on the mouse milk fat globule. See also Ceriani et al. J. Natl. ^• Cancer Inst. (1978) 61:747-751; Ceriani et al., Use Of Mammary Epithelial Antigens As Markers In Mammary Neoplasia In: Boelsma E., Rum e Ph. eds. Tumor Markers: Impact and

Prospects, New York, Elsevier-North Holland, 1977:101-106. Ceriani et al. , J. Nuc. Med. (1976) 17:567 report the use of whole polyclonal antisera for localizing mouse mammary epithelial carcinomas in mice. Peterson et al., Abstract, 32nd Annual Meeting of the Tissue Culture Association,

Washington, D.C., June 7-19, 1981 report monoclonal anti¬ bodies to differentiation antigens of normal human mammary epithelial cells.

SUMMARY OF THE INVENTION Labeled antibody fragments of less than about

80,000 daltons specific for mammary epithelial cells are employed for site directed localization or imaging of mam¬ mary neoplastic tissue, whether the tissue is associated with the breast or other organ. Particularly, poly- or monoclonal Fab fragments labelled with radionuclides speci¬ fic for normal mammary epithelial cell surface antigens are introduced into a host for selective specific binding to mammary epithelial tissue, providing a high ratio of speci-

fie binding to background, while allowing for rapid elimina tion of the radionuclide labelled fragment. By employing a physiologically acceptable radionuclide other than the labe in conjunction with the labelled antibody fragment, back- ground values can be obtained which permit accurate detec¬ tion of neoplastic mammary tissue.

DESCRIPTION OF THE SPECIFIC EMBODIMENTS The subject invention provides methods and composi tions for specifically localizing or imaging mammary tissue, particularly detecting neoplastic mammary tissue, regardless of the aggregation site of the mammary epithelial cells. High specificity is achieved by obtaining antibodies to antigenic sites of normal mammary epithelial cells, where the antibodies bind to both normal and neoplastic cells. The subject invention permits localization, not only of carcinomas located at the breast but also metastatic carcinomas from a primary breast carcinoma of origin. Thus, •even after a breast cancer is removed, where cancerous colonies have developed from the primary breast carcinoma, these will be readily imageable.

The hosts of interest for mammary carcinoma detec¬ tion are mammals, which includes research animals, domestic animals, including pets and farm animals, as well as humans. Antibodies may be obtained by employing as an immunogen delipidated milk fat globules containing mammary surface epithelial antigens. Alternatively, whole cells, lysates, surface antigens or determinant portions of surface antigens may be employed as a source of normal surface antigens to serve as immunogens. Depending upon whether one can use allogeneic or xenogeneic antibody fragments, various tech¬ niques can be employed for producing the antibody fragments.

Where xenogeneic antibodies may be used, milk is obtained from a nursing host, the cream fraction isolated, milk fat globules in the cream fraction freed of the fat employing organic solvents e.g. chloroform and ether, and the residue lyophilized. The lyophilized powder may then be used as an immunogen in a host different from the host source of the milk.

Where other than delipidated fat globules are employed, surface antigens or fragments thereof obtained from whole cells may be used for the preparation of poly- clonal or monoclonal antibodies. Surface antigens may be isolated from defatted fat globules or by lysing mammary epithelial cells and using an affinity column, employing antibodies specific for the mammary epithelial surface antigens for separation of the surface antigen. Individual surface antigens may be freed from the affinity column and separated by various techniques, such as chromatography, electrophoresis, isopycnic separation, gradient density, or other conventional techniques.

The individual surface antigen may then be employed as an immunogen or the protein partially digested and the resulting fragments obtained and employed for affinity chromatography using the antibodies described above to isolate peptides providing the desired determinant sites. These antigens or fragments thereof may be injected into a host other than the host source of the antigen or may be combined, either covalently or noncovalently, to an immuno¬ gen foreign to the host source and then injected into the host source for production of antisera. Where the host is human, normally the technique employed will be for prepara¬ tion of monoclonal antibodies. For monoclonal antibodies, a wide variety of conventional techniques may be employed, primarily based on the seminal work of Kohler and Milstein. For monoclonal antibodies, usually other than the whole cell will be em¬ ployed, normally either the antigen or a fragment having a desired determinant, site. With other than humans, the monoclonal antibodies may be obtained by injecting the antigen, immunogenic fragment of an antigen or an haptenic fragment bound to an immunogen, into a host different from the host source of the antigen with an appropriate adjuvant, repeating the injection about two to four weeks later.

This is followed by isolation of the spleen and fusing of the splenocytes with an appropriate fusion partner e.g. a myeloma or lymphoma cell, from the same or different

tm

species as the immunized host. The resulting hybridomas ma then be screened for antibodies which bind to the deter¬ minant site(s) of interest. Besides splenocytes, peripheral blood cells or lymphocytes may be immunized in vitro and used as the partner for fusion with the immortalized cell line. The immunoglobulin will then be harvested in conven¬ tional ways and treated to provide specific binding fragments.

Where polyclonal antisera are involved, the IgG will be purified followed by digestion to provide a frag¬ ment. Various fragments may be obtained by conventional techniques, such as Fab, F(ab) ₂ , Fd or Fv. Various pep- tidases have been reported which cleave antibodies at speci¬ fic sites or controlled proteolytic degradation can be employed. Each of these fractions may have advantages in providing the desired degree of affinity, rate of elimina¬ tion, ^' acceptable degree of immunogenicity, or the like. •Various proteases may be employed for digesting the antisera and removing undesired fragments to leave the binding por- tion, normally having a single binding site.

The resulting digested antibody fragment will be purified further, either before or after labelling, normally before labelling, by absorption with host source cells other than the mammary epithelial cells, to remove cross-reacting antibody fragments. Depending upon the particular host, red blood cells, liver homogenates, fetal fibroblasts and var¬ ious cell lines e.g. HeLa, colorectal carcinoma, etc., may be employed to further insure the specificity of the anti¬ body fragments which are used. During the absorption and removal of cross react¬ ing antibodies in polyclonal antisera, various components from the cells may be secreted or be leached into the anti¬ serum composition. In order to ensure the substantial absence of such impurities, it will normally be desirable to further purify the antibody fragments using an affinity chromatography column, employing the complementary antigen or antibody specific for immunoglobulins from the host source of the immunoglobulins. The antibody fragment compo-

sition is combined with the complementary binding member insolubilized to a support to provide an inεolubilized phase pair member and the specific antibody fragments are then bound to the insolubilized phase. The insolubilized phase is then washed thoroughly to remove any non-specifically bound components and the purified antibody fragments released using a solution such as 2M sodium isocyanate, followed by rapid dialysis.

A wide variety of labels may be employed, depend- ing upon the purpose and function of the labeling, e.g. making the neoplastic tissue visualizable. Particularly, radionuclides may by employed for therapy, in vivo diagnosis, physiological studies, or the like. Various radionuclides may be employed, commonly radioactive iodine e.g. ¹³¹ I ¹²⁵ I, ¹²³ I, and ^99m Tc. Conventional labeling techniques are employed, conveniently the chloramine-T method for iodine of Greenwood and Hunter, Biochem. J. (1963) 89:114-123 or lactoperoxidase. Other labels include fluorescers which may be covalently conjugated through carboxyl groups, isothiocyanate, thio groups, activated olefins, or the like. Other labels may also be useful as imaging detection techniques are evolved.

The resulting labelled composition may then be injected intravenously into the host in an appropriate physiologically acceptable vehicle. Conveniently, phosphate buffered saline, saline, Ringer's solution or other aqueous buffered medium may be employed. Generally, about 5 to 30μCi, more usually about 10 to 25μCi, will be injected per kilogram of host where a radionuclide is employed for imaging of a carcinoma. For radioimaging,. desirably a different radionuclide will be injected into the host to provide for a background control so that the concentration of the labeled antibody fragment is determined by the ratio of the emission from the radionuclide label divided by the emission from the background radionuclide. Usually, the scanning of the distribution of the radionuclide will be made from about 6 to 48 hrs. after injection. Various

1ϋl

instruments may be used for scanning of the radionuclide distribution.

The subject compositions employing the antibody fragments may be formulated and supplied labelled with a moderately long half-lived radionuclide e.g. 131I or sup¬ plied unlabelled to be labelled prior to use. The formu¬ lated product may be provided as a concentrate, at concen¬ trations 1.5 to 10 times the concentration to be employed _^ for administration or may be provided at the desired con- centration. Depending on the mammal to be imaged, the dosage will vary. Since humans will be the most commonly treated, the indicated dosages are primarily for humans.

Compositions of the labelled antibody fragment may be preprepared or be prepared for use shortly before injec- tion, e.g. intravenously . These compositions will contain from about 10 to 300μg protein/ml of the labelled antibody

4 fragment and have from about 100 to 10 μCi/ml. Usually, about 1 to 3 ml of the composition will be injected. Prior to admi .ni.strati.on, c ^* onveniently about 200 to 5x103μCi of

QQγn Tc-pertechnetate may be added to the composition prior to injection. The technecium provides for a background level of radiation which can be distinguished from the emission from radioactive iodine so as to give a measured natural variation in distribution of the radionuclide. Of the materials which may be included in the composition are preservatives, such as benzyl alcohol, buffers e.g. phos¬ phate, salt (NaCl), inert protein e.g. serum albumins. The concentration of the buffer will generally by 0.03 to 0.07M, while the concentration of the salt will generally be 0.11 to 0.18M, while the amounts of the other materials may be varied widely, the inert proteins generally range from about 0.1 to 2 weight percent and the additives generally ranging from about 0.0001 to about 0.1 weight percent, when present. The pH will generally be in the range of about 7 to 7.8. In order to further demonstrate the subject inven¬ tion, the following examples are offered by way of illustra¬ tion and not by way of limitation.

OMPI

Antisera were prepared essentially as described in Ceriani and Peterson, supra. After an injection of oxytocin, mouse milk was collected by suction from nursing mice separated from their pups overnight. The cream frac- tion was collected after centrifugation (13,000 g) and washed three times with phosphate-buffered saline (PBS). The final cream fraction consisting of the milk fat globules (MFG) was suspended in water and washed three times with chloroform and three times with ether to remove the fat and then lyophilized. New Zealand white rabbits were immunized with a 5-mg injection of this defatted MFG membrane prepara¬ tion followed by an injection 3 weeks later of 3 g. Serum was collected 10 days after the second injection and the serum globulins were precipitated by 40% saturated ammonium sulfate. The globulins were redissolved in PBS, dialysed against PBS, and then heated at 60C for 30 min. to destroy complement before sodium azide was added. The globulins -were stored frozen at -20°C.

For preparation of the Fab fragments, the IgG was first purified from the stored globulin fraction by gel filtration chromatography on a 27x2.5-cm diameter Sephadex G-200 column with 0.1m Tris, pH 8.0, 0.15 NaCl, 0.1% sodium azide. One hundred 2 ml fractions were collected at room temperature during the five hour elution period and the peak fractions determined by optical density at 280 μm. The IgG came off the column as a single peak separated from and after the void volume. The fractions of this peak were pooled.

The digestion of IgG with papain was a modifica- tion of the method of Porter, Biochem. J. (1959) 73:119-127. After its protein concentration was determined (Lowry) the IgG fraction was dialyzed against 0.1 m phosphate, pH 7.5, 0.002 m ethylenediaminotetraacetic acid diluted by an appro¬ priate factor necessary to retain these molarities after later concentrating the IgG to 15 mg/ml. (All dialyses were done at 4°C.) After dialysis, the IgG solution in the tubing was concentrated by dehydration until the appropriate concentration was achieved. It was brought to 0.01 m

cysteine, and then 1 g of twice recryεtallized papain (Sigma Corp.) was added for every 100 mg of IgG present. This mixture was incubated for 2 hours at 37°C, then dialyzed against frequent changes of water, which inacti- vated the enzyme and coincidentally caused crystallization of many of the Fc fragments. The dialysis buffer was then changed to 0.01 m acetate, pH5.5. After dialysis, the precipitate was removed and the supernatant was poured on a. 37 ml carboxymethylcellulose column, 1 cm in diameter, at room temperature. Fifteen 5 ml fractions were eluted with

0.01 m acetate buffer, pH5.5 at a rate of one fraction every 23 in. A 400 ml gradient from 0.01 m to 0.9 m acetate was then applied to the column and 5 ml fractions again col¬ lected at the same rate. The fractions were assayed for optical density at 280 μm, and the first peak (Fab) was sterilized by filtration and stored at 4°C. Protein concen¬ trations were determined by the method of Lowry. Fifteen mg Fab were recovered from 60 mg IgG, giving 45% of the theoretical yield. Iodination was achieved by a modification of the chloramine-T method of Greenwood and Hunter, Biochem. J.

(1963) 89:114-123. A reaction mixture of 30 μg Fab in 80 μlθ.5 m phosphate, pH 7.4, was made up and 1 Ci 125I added.

When ¹³¹ I was used instead of ¹²⁵ I, 5 Ci of ¹³¹ I and 190 μg Fab were used. The reaction was initiated with 50 μl of chloramine-T (4 mg/ml) and stopped after 30 seconds with 80 μl of sodium metabisulfite (4.8 mg/ml) followed by 100 μl KI

(10 mg/ml). The reaction mixture was then layered onto a 5 ml Sephadex G 25 column packed in a 10 ml disposable plastic pipette that was equilibrated with PBS containing 0.1% bovine serum albumin (BSA); 0.3 ml fractions were collected and monitored on a Packard Tri-Carb Scintillation Counter,

Model 3002, and the peak fractions pooled. Forty to 60% of the iodine was usually recovered in the protein peak. Cross-reacting Fab fragments were then removed by serial absorptions with mouse red blood cells, washed mouse liver homogenates, and mouse fetal fibroblasts. The anti-

MME (Fab) absorbed in this way bound specifically to mouse

mammary epithelial cells and not to cells from other tissues of the mouse as revealed by immunofluorescence techniques, as was the case for the intact anti-MME. The labeled, absorbed anti-MME (Fab) was stored at 4°C and used within 10 days. Nonspecific Fab of a nonimmunized rabbit was pre¬ pared, labeled and absorbed in the same manner as the speci¬ fic anti-MME (Fab).

Monoclonal antibodies specific for normal mammary, epithelial cell surface antigens may be prepared as follows. Delipidated human milk fat globules (HMFG) or specific purified components are injected into BALB/c mice and receive at least one booster from two to three week inter¬ vals. Three to four days after the last booster injection, the immunized mouse spleen cells are fused with mouse myeloma cells in the presence of PEG6000 and then plated in 600 microwells plus feeder layers. HAT medium is added and the supernatants from HAT resistant clones screened for binding to insolubilized HMFG. HMFG positive clones are screened by adding the supernatants to microtiter wells to which are bound cell membranes from the following cell types: HeLa, HT-29, human mammary fibroblasts, and human lymphoma (Bristol-8). Antibodies which specifically bind to the above membranes may be detected with radio- or other labelled anti(mouse Ig). The desired clones are those that bound to HMFG, but not to membranes of HeLa, colon carci¬ noma, human fibroblasts and human lymphoma cell membranes. The monoclonal antibodies may then be produced by injecting the selected hybridoma cells into BALB/c mice to produce ascites fluid or growing the cells and harvesting the media in which the hybridomas are grown. The monoclonal anti¬ bodies are then purified with affinity columns containing anti(mouse Ig) on the solid phase.

Tumor Lines

The two transplantable mammary tumors, 3910-30 (Ceriani et al, (1978) supra) and HOG (Medina and DeOme, J. Natl. Cancer Inst. (1970) 45:353-363) were maintained in BALB/c mice by subcutaneous transplantation. Both were

spontaneous mammary tumors that arose in BALB/c and have shown no mouse mammary tumor virus particles upon electron microscopic examination. The two lines that were non- mammary lines (Ceriani et al, (1976) supra) arose in C57/BL mice and were maintained in that strain.

Tissue Distribution Studies

In order to study the tissue distribution of absorbed anti-MME (Fab), mammary-tumor-bearing mice were injected intravenously with 0.25 μCi of the 125I labeled preparation in 0.1 ml PBS. The mice were dissected after 24 hours and samples of blood, urine, and tumor taken, along with the brain, liver, lung, kidneys, and the muscle and bone of one hind leg. The tissue pieces were weighed and counted on a Packard Scintillation Counter. Values of cpm per gram obtained by similar injections of 125I-labeled BSA were subtracted to account for blood content of each tissue.

Imaging

Mice to be imaged were given 1 ml KI in 0.1 ml PBS to block 131I uptake by the thyroid, followed by an mtra-

131 venous injection of 70 μCi of I-labeled anti-MME (Fab).

Approximately 23 hr. later, they were given 500 μCi of

^c-pertechnetate under the skin of the scalp. Images were obtained 60 to 90 min. later on the germanium camera after the mice were anesthetized with sodium nembutal. The camera's structure, parameters, and performance have been described by Kaufman et al., Invest. Radiol (1978)

13:223-232; Kaufman et al., Information Processing in

Medical Imaging, (1979), Vol. 88, Paris, Editions INSERM

(1980); 153-172. For this work, a 7-mm pinhole aperture was used. The 131I image was obtained with a 353-366 KeV win¬ dow, and a separate ^Tc image was obtained with a 136-145 KeV window.

An on-line PDP 11-34 computer controlled the data acquisition and displayed the images on black and white and color television monitors. The 131I images could be normal¬ ized to the "'TC images by a program that divided the 131.

counts by the xc counts for each imaging element. Syringes filled with ^ c were used as fiducial markers to reference the image to actual photographs of the animals. Mice carrying transplanted mammary tumors were injected intravenously with 0.25 μCi of labeled, absorbed anti-MME(Fab) . Twenty-four hr. later, they were sacrificed and the amount of label in various tissues was determined.

TABLE I

Percentage of .administered dose (cpm/g tissue) ¹

Anti-MME Nonspecific Anti-MME Nonspecific

(Fab) Fab (Y(G) ² γG ³

Liver 0.07±0.022 0.21±0.02 0.81±0.25 0.80±0.11

Brain 0.006±0.004 0.02±0.004 0.033±0.007 0.028±0.002

Kidney 1.6±0.4 1.30±0.18 0.00 0.66±0.45

Muscle 0.08±0.01 0.10±0.02 0.38±0.14 0.35±0.007

Lung 0.18±0.02 0.17±0.04 — —

Tumor 0.87±0.26 0.3±0.004 2.54±0.23 1.44±0.04

¹ Radioactivity of the circulating blood in each tissue has been subtracted.

² Each value is the mean of five mice with standard error of mean for Fab and three mice for γG. ³ Data taken from Cerriani et al. (1979).

Most tissue levels were lower than those found previously with whole gamma globulin. In addition, the entire blood pool of the animals was calculated to contain only 1% of the injected dose at 24 hrs. Time-course studies showed tissue levels to drop rapidly toward background levels by 2 days. This was in marked contrast to the radio- iodinated gamma globulin preparation, which showed increas¬ ing accumulation of label in the liver for at least three days. The kinetics of appearance of Fab in the urine indi- cated that by 19 hr. , 99% of it had been excreted. These results agree closely with the work on Fab catabolism done by Spiegelberg and Weigle, I. Exp. Med. (1965) 121:323-338.

The labeled Fab appeared to pass intact through the kidneys into the urine, since 96% of the 125I recovered in the urine was associated with protein. In polyaery1amide gel electrophoresis, the labeled Fab preparations were found to consist of a molecule of 50,000 daltons, the molecular weight expected for a rabbit Fab fragment, along with its apparent dimer of about 100,000 daltons. The labeled mater¬ ial in urine had a nearly identical electrophoretic profile.. Treatment of both the labeled Fab and urine with the reduc- ing agent β-mercaptoethanol yielded subunits of 26,000 daltons, the size expected for the polypeptides that make up the Fab fragment.

At 24 hours after injection of the labeled anti- MME (Fab) there was a 5- to 10-fold greater accumulation of label in the mammary tumor as compared with liver, brain, lung, and muscle tissue.

When labeled nonspecific Fab was injected into the tumor-bearing mice there was still some accumulation of label in the tumor, but considerably less than with the specific antiserum. This can be explained by the tendency of tumors to sequester proteins nonspecifically, one reason why tumor imaging with labeled serum albumin has had some success. Taking into account this nonspecific sequestering of proteins, the labeled anti-MME (Fab) still showed a threefold to sevenfold accumulation in mammary tumors, as compared with liver, lung, brain, and muscle.

Mice similar to those used for tissue distribution studies were given intravenous injections of 70 μCi of

131 I-labeled, absorbed anti-MME (Fab), 24 hours before imaging. They were also given 500 μCi of . C- pertechnetate under the skin of the scalp one hour before imaging.

The iodine images from the germanium camera typically contained 5,000-10,000 counts, accumulated over a 45-60 min. period. Technecium images, containing about 100,000 counts, were obtained over a 5-10 in. period.

Images obtained were present on color or black- and-white television screens by the camera's computer. The

computer employed a scale of 64 colors or greys with 64 different shades to indicate intensity. The point with the most counts in each picture was arbitrarily set to 64 and all other points normalized by that value. This resulted in the most counts intense area in each image having the same visual intensity as that in any other image. A printed matrix gave the actual counts for each imaging element, enabling the quantitation and comparison of density of counts over different areas of the image. The 99_π_T_____c images mainly represent the extracellu¬ lar and intravascular spaces of the animal with the excep¬ tion of areas of physiological accumulation, such as the stomach and bladder. The prominent area of concentration at the midportion of the mice is most likely the blood-filled liver and spleen, and accumulation in the stomach. Some¬ times, a concentration of "'Tc also remained at the site of its injection. The 131I image of nonmammary tumor-bearing animals showed relatively even distribution of the label throughout the animal's body and over the tumor with slightly concen¬ trated areas over the liver, and sometimes over the bladder as was also the case with nonspecific Fab used in mice with mammary tumors. The 131I images of mammary tumor-bearing animals injected with anti-MME (Fab) showed concentration of the label over the tumor at levels distinctly higher than those of other tissue of the animal.

A normalized image was generated consisting of the ratios of I to ^c at each imaging element. The nor¬ malization procedure allows an increase in the specificity of the 131I imaging by compensating for differences in extracellular and intravascular spaces in each individual mouse, with the reservation that areas of physiological accumulation, such as the stomach and bladder are not in¬ cluded when comparing the density of label in different areas of the mouse. In the normalized ( 1/ ^c) images there were usually a few hot spots outside of the body of the mice. These are artifacts of the normalization process that occur as a result of the very low counts in these areas

and their resulting poor statistics. In the normalized ( 1311/99mTc) images, the mice carrying mammary tumors had a clearly defined concentration of label over their tumors.

In the case of the mouse carrying the HOG mammary tumor, th localization was better in this 1/ "' C image than the

131 I image. In the mice carrying nonmammary tumors or when nonspecific Fab was used, the density of label over the tumor in the normalized image did not appear significantly^ greater than over comparable nontumor areas of the mouse.

The visual ambiguities of the 1-64 scale were eliminated by using a printed matrix of the counts in each imaging element, to quantitate results. Projection of the photograph of the mouse on this matrix permitted determining the average counts per imaging element in the area of the tumor and over a comparable area on the opposite side of the mouse. From the average counts per imaging element the localization ratio (tumor/opposite side) was determined for the ^99m Tc, ¹³¹ I and ¹³¹ I/ ^99m Tc images.

TABLE II Comparison of Localization of 131I-Labeled

Anti-MME (Fab) Fragments in Transplanted

Mammary and Nonmammary Tumors in Mice

*

Localization Ratio

^99m Tc ¹³¹ ι ¹³¹ I/ ^99m Tc ratio ratio ratio

Mammary tumors

HOG 0.64 • 2.6 4.1

3910-30 1.2 4.6 3.9

3910-30 (Non 2.0 0.9 0.4 specific Fab)

Nonmammary tumors

Melanoma (B-16) 0.83 1.05 1.26

Lewis lung carcinoma 0.55 0.56 1.0

The localization ratio is found by dividing the average counts per imaging element of the germanium camera image.

over the tumor by the average counts over a comparable area on the opposite side of the mouse.

The difference in extracellular and intravascular spaces in the different mice can be seen in the variation in ^Cc ratios with values ranging from 0.5 to 2.0. On the contrary, there is a fourfold greater accumulation of 131I in the mammary tumor as indicated by the normalized localiza tion ratio ( 1/ ^c) of 4.0 for both mammary tumors. In. the case of the mouse carrying the HOG mammary tumor, the normalization procedure quantitatively improved the imaging. When 131I-labeled nonspecific Fab is used in mammary tumor- bearing mice, the localization ratios for 131I and

131 1/99mTc are close to 1.0 or even less indicating no non- speci .fi.c accumulation of 131I-Fab m the tumor. It is evident from the above results, that by using antibody fragments specific for normal mammary epithelial cells, a rapid accurate imaging of tumors can be Obtained regardless of the site of the tumor. The fragments are rapidly eliminated, so that there are minimal effects from the radiation resulting from the radionuclide label.

Despite the rapid elimination, the antibody fragments still retain sufficient binding affinity to rapidly accumulate in the host by primarily specific binding to normal mammary epithelial tissue to allow for sharp discrimination between mammary epithelial tissue and other tissues present in the host. Thus, one can not only detect the presence of a tumor in the mammary gland, but also etastatic tumors which may have resulted from a mammary tumor.

By employing labelled antibody fragments which are specific for normal mammary epithelial cell surface antigen determinant sites localization is achieved wherever neoplastic mammary tissue exists in the host. Thus, not only are tumors located throughout the host, but the primary carcinoma of origin is also determined, where the primary carcinoma of origin was a mammary carcinoma.

Although the foregoing invention has been described in some detail by way of illustration and example for purposes of clarity of understanding, it will be obvious

that certain changes and modifications may be practiced within the scope of the appended claims.