NO178341B - Analogous Process for Preparation of Therapeutically Active Anthracycline Derivatives - Google Patents

Description

The present invention relates to an analogue method for the production of hitherto unknown, therapeutically active anthracycline compounds which are prodrugs. A tumor-specific antibody-enzyme conjugate that binds to the tumor cells is administered, and further administration of the prodrug converts this in the tumor area in the presence of the antibody-bound enzyme into an active cytotoxic drug. The antibody-enzyme conjugates and prodrugs produced according to the invention solve many of the disadvantages of the antibody-mediated drug delivery systems that are currently used for the treatment of cancer and other tumors.

The invention thus relates to an analogue method for the preparation of a therapeutically active anthracycline derivative of the formula

in which

R is the group

or

R<1> is H, and R<3> is OH or OCH3, or

R<1> is OH, and R<3> is OCH3, and

R<2> is H or OH,

which method is characterized by the fact that a compound of the formula

in which R<1> and R3 have the meaning given above, or an acid addition salt thereof, is reacted with a compound of the formula or

wherein R<2> has the above meaning, or an acylating equivalent thereof.

Use of iramunoconjugates for the selective delivery of cytotoxic agents to tumor cells in treatment

ling of cancer is known in the art. Delivery of cytotoxic agents to the tumor areas is highly desirable, as systemic administration of these agents often results in the destruction of normal cells in the body, as well as the tumor cells that are sought to be destroyed. According to the currently used antitumor drug delivery systems, a cytotoxic agent is conjugated with a tumor-specific antibody, forming an iramunoconjugate that binds to the tumor cells and thereby "delivers" the cytotoxic agent in the tumor area. The immunoconjugates used in these targeted systems include antibody-drug conjugates [see e.g. R. W. Baldwin et al. "Monoclonal Antibodies For Cancer Treatment", Lancet, pp. 603-05 (March 15, 1986)] and antibody-toxin conjugates [see, e.g. P.E. Thorpe "Antibody Carriers Of Cytotoxic Agents In Cancer Therapy: A Review"

in Monoclonal Antibodies '84: Biological and Clinical

Applications, A. Pinchere et al. (ed.), pp. 475-506 (1985) ].

Both polyclonal antibodies and monoclonal antibodies have been used in these immunoconjugates [see e.g. K. Ohkawa et al. "Selective In Vitro And In Vivo Growth Inhibition Against Human Yolk Sac Tumor Cell Lines By Purified Antibody Against Human a-Fetoprotein Conjugated With Mitomycin C Via Human Serum Albumin", Cancer Immunol. Immunother., 23, pp. 81-86 (1986) and G.F. Rowland et al. "Drug Localization And Growth Inhibition Studies Of Vindesine-Monoclonal Anti-CEA Conjugates In A Human Tumor Xenograft", Cancer Immunol. Immunother., 21, pp. 183-87

(1986) ]. The drugs used in these immunoconjugates include daunomycin (see, for example, J. Gallego et al. "Preparation Of Four Daunomycin-Monoclonal Antibody 791T/36 Conjugates With Anti-Tumor Activity", Int. J. Cancer, 33, p. 737 -44

(1984) and R. Arnon et al. "In Vitro And In Vivo Efficacy Of Conjugates Of Daunomycin With Anti-Tumor Antibodies", Immunological Rev., 62, pp. 5-27 (1982)], methotrexate [N. Endo et al. "In Vitro Cytoxicity Of A Human Serum Albumin-Mediated Conjugate Of Methotrexate With Anti-MM46 Monoclonal Antibody", Cancer Research, 47, pp. 1076-80

(1987) ], mitomycin C [K. Ohkawa et al., supra], and vindesin [G.F. Rowland et al., supra]. The toxins used in the antibody-toxin conjugates include bacterial toxins such as diphtheria toxin and plant toxins such as ricin [see e.g. F. L. Moolten et al. "Antibodies Conjugated to Patent Cytotoxins As Specific Antitumor Agents", Immunol. Rev., 62, pp. 47-73 (1982)].

Despite the extent of research directed towards the use of immunoconjugates for therapeutic purposes, several limitations associated with these delivery methods have become apparent [see e.g. M. J. Embleton "Targeting Of Anti-Cancer Therapeutic Agents By Monoclonal Antibodies", Biochemical Society Transactions, 14, pp. 393-395 (615th Meeting, Belfast 1986)]. First, the large amount of drug that needs to be delivered to the tumor cell to induce destruction of the cell is often unattainable due to limitations due to the number of tumor-associated antigens on the surface of the cells, and the number of drug molecules that can binds to a given antibody molecule. This limitation has led to the use of more powerful cytotoxic agents such as plant toxins in these conjugates, and to the development of polymer-bound antibody-drug conjugates with very high drug-multiplicity ratios [see e.g. P.E. Thorpe, supra, pp. 475-506, and R.W. Baldwin et al. "Design And Therapeutic Evaluation of Monoclonal Antibody 791T/36 - Methotrexate Conjugates" in Monoclonal Antibodies and Cancer Therapy, pp. 215-31 (Alan R. Liss, Inc. 1985)]. However, even with high drug loading ratios or using potent toxins, many immunoconjugates still exhibit sub-optimal cytotoxic activity and are unable to elicit complete killing at doses where all available antigenic sites are saturated.

Secondly, it is obvious that the cytotoxic activity of an immunoconjugate often depends on the uptake thereof in the tumor cell mediated by the antibody component of the conjugate [see e.g. J.M. Lambert et al. "Purified Immunotoxins That Are Reactive With Human Lymphoid Cells", J. Biol. Chem., 260 (no. 22, pp. 12035-12041 (1985)]. This internalization is crucial when an antibody-drug conjugate is used where the drug has an intracellular area of action, or when antibody-toxin conjugates are used. The main part of tumor- associated antigens, and thus antibody-drug or antibody-toxin conjugates bound to these antigens, are not internalized, however. The conjugates that are internalized are often transported to the cell's lysosome where the drug or toxin is degraded [see E.S. Vitetta et al., Science, 238 , pp. 1098-1104 (1987)].Although an antibody-drug or antibody-toxin conjugate may exhibit excellent tumor binding properties, the conjugate may nevertheless exhibit limited cytotoxic utility due to an inability to reach its site of action within the cell.

It is also well known that tumor cell populations are often heterogeneous with respect to antigen expression [see e.g. A. P. Albino et al. "Heterogeneity In Surface Antigen and Glycoprotein Expression of Cell Lines Derived From Different Melanoma Metastases of The Same Patient", J. Exp. Med., 154, pp. 1764-78 (1981)]. In addition, it has been shown that antigen-positive tumor cells can give rise to antigen-negative offspring [see e.g. M. Yeh et al. "Clonal Variation for Expression of a Human Melanoma Antigen Defined by a Monoclonal Antibody", J. Immunol., 126 (No. 4), pp. 1312-17 (1981)]. In any population of tumor cells, there will thus be a certain number of cells that do not exhibit the antigen for which a particular immunoconjugate exhibits specificity. The immunoconjugate will therefore not be able to bind to these cells and mediate their destruction.

As a result of these disadvantages, the currently used antitumor drug or toxin delivery systems have limited success, particularly when applied to in vivo treatment.

In addition to the immunoconjugates described above, antibody-enzyme conjugates have been investigated in vitro in combination with another non-targeted enzyme to convert iodide or arsphenamine to their toxic forms, in order to enhance antibody-mediated cytotoxicity [see e.g. C. W. Parker et al. "Enzymatic Activation and Trapping of Luminol-Substituted Peptides and Proteins. A Possible Means of Amplifying the Cytotoxicity of Anti-Tumor Antibodies", Proe. Nati. Acad. Pollock. USA, 72 (No. 1), pp. 338-42 (1975), and G.W. Philpott et al. "Affinity Cytotoxicity of Tumor Cells With Antibody-Glucose Oxidase Conjugates, Peroxidase, and Arsphenamine", Cancer Research, 34, pp. 2159-64 (1974)].

According to these in vitro studies, the enzyme gluco-oxidase is bound to an antibody and used in combination with a non-targeted peroxidase enzyme to convert iodide or arsphenamine to, respectively, cytotoxic iodine or arsenic. This method therefore not only requires gluco-oxidase to be targeted against tumor cells with antibody, but also the presence in the tumor area of two other non-targeted agents. The probability that all three agents in vivo will be found in the tumor area at the same time is small, and it is therefore unlikely that this method will be of therapeutic importance.

In Canadian patent document 1,216,791, conjugation to an antibody of an enzyme which can release ammonium ions from the substrate is described. Ammonium ions are released to enhance the cytotoxic effect of immunotoxins targeted at the tumor area.

Finally, European patent application no. 84302218.7 deals with a method for treating a diseased cell population such as a tumor, in which method an antibody is used to direct a non-metabolized antigen against the tumor cells. The antigen accumulates within at least a percentage of the tumor cells which are then lysed releasing the antigen in a ubiquitous fibronectin capture matrix formed in the tumor area. At this point, the method according to this invention administers an iodine-containing ligand which is specific for, and which will bind to, the antigen bound to the matrix. The cytotoxic iodine then reacts to destroy the tumor cells at this site. Many embodiments are described in this application, and one of these proposes the use of an antibody-enzyme conjugate to target the enzyme to the tumor area, and the addition of a non-lethal substrate that the enzyme can convert into a cytotoxic material [see indicated European patent application page 34- 35]. However, nowhere in the application is it described how this embodiment is to be implemented. Similarly, Hellstrom et al. "Antibodies For Drug Delivery" in Controlled Drug Delivery (2nd ed.), Robinson and Lee (eds.), p. 639 (1987), that "[d]rugs which would be non-toxic until 'activated' by an agent (e.g. an enzyme) localized to tumor may be considered as another approach

So far, however, no one has described or indicated how the present method can be carried out, or has attempted to carry out this drug-targeting method.

The above-mentioned problems are remedied by delivery of cytotoxic agents to tumor cells by combined use of antibody-enzyme conjugates and the produced prodrugs. An enzyme that can convert a mildly cytotoxic or a non-cytotoxic prodrug into an active cytotoxic drug is conjugated with a tumor-specific antibody. This antibody-enzyme conjugate is administered to a tumor-bearing mammalian host and, as a result of the antibody specificity, binds to the surface of these tumor cells which exhibit the tumor antigen against which the antibody's specificity is directed. Prodrugs produced according to the invention are then administered to the host and converted in the tumor area by the action of the antibody-bound enzyme into a more active cytotoxic drug.

The invention is explained in more detail below in connection with the drawings in which

Figure 1 depicts the strategy for activation of prodrugs by tumor cells that bind antibody-enzyme conjugates. Figure 2 depicts the chemical structure of an adriamycin prodrug ("APO") and its preparation from adriamycin.

Figure 3 is a comparative graphical presentation of the percentage of adriamycin release versus time when reacting APO with A: free penicillin V amidase enzyme or o and D: the L6-PVA conjugate - in total or 10 and 100 µg protein/ml. The progress of the reaction was recorded with HPLC.

Figure 4 depicts a comparison between the binding of the monoclonal antibody L6 and the L6-PVA and 1F5-PVA conjugates to H2981 tumor cells.

Figure 5 is a comparative graphic representation of the percentage inhibition of <3>H-thymidine incorporation into DNA of H2981 tumor cells treated with o : adriamycin (ADM), • : APO, A : L6-PVA+APO or ▲: 1F 5 -P VA+APO. The curve depicts the observed increase in cytotoxic activity when the tumor cells were penetrated with the L6-PVA conjugate, followed by APO treatment, compared to the activity observed when treated with APO alone.

Figure 6 depicts a comparison between the binding of the monoclonal antibodies L6 and 1F5 and the L6-PVA and 1F5-PVA conjugates to Daudi lymphoma cells.

Figure 7 is a comparative graphical representation of the percentage inhibition of <3>H-thymidine incorporation into DNA of Daudi lymphoma cells treated with • : ADM, o : APO. ■ : L6-PVA+AP0 or □ : 1F5-PVA+AP0. The curve depicts the observed increase in cytotoxic activity when the tumor cells were penetrated with the 1F5-PVA conjugate, followed by APO treatment, compared to the cytotoxic effect observed after treatment of the cells with APO alone.

Antibody-enzyme conjugate administered to a tumor-bearing mammalian host consists of a tumor-specific antibody bound to an enzyme that can convert a prodrug that is less cytotoxic to tumor cells than the parent drug into the more active parent drug. Upon administration to a host, the anti-substance f component of the conjugate, which component is reactive to an antigen present on the tumor cells, carries the conjugate to the tumor site and binds to the tumor cells. The antibody can therefore be regarded as a means of delivering the enzyme to the tumor area. A prodrug that is a substrate for the enzyme is then administered to the host and converted by the enzyme in the tumor area into an active cytotoxic drug. The drug is thus activated extracellularly and can diffuse into all tumor cells in this area, i.e. the cells that carry the specific tumor antigen for which the antibody in the conjugate shows specificity, and to which the antibody is bound, as well as the cells that are negative with regard to to this antigen, but is nevertheless present in the tumor area (see Fig. 1). The problem of tumor-antigen heterogeneity and the requirement for antigen/conjugate internalization associated with conventional immuno-conjugate drug delivery methods is therefore resolved.

As the drug is not required to bind directly to the antibody and thereby limit the amount of drug that can be delivered, the general problem of drug potency in the tumor area does not arise. In reality, the delivery method increases the number of active drug molecules found in the tumor area, as the antibody-bound enzyme in the conjugate can perform numerous substrate conversions as the prodrugs are repeatedly converted into active drug. In addition, the delivery method is able to specifically release the active drug in the tumor area as opposed to release in other tissues. This is the case because the enzyme concentration in the tumor area is higher than the concentration in other tissues as a result of coating the tumor cells with the antibody-enzyme conjugate.

The antibody component of such immunoconjugates includes any antibody that specifically binds to a tumor-associated antigen. Examples of such antibodies include, but are not limited to, such antibodies that specifically bind to antigens found on carcinoma, melanoma, lymphoma and sarcoma in bone and soft tissue, as well as other tumors. Antibodies which remain bound to the cell surface for longer periods of time, or which internalize very slowly, are preferred. These antibodies can be polyclonal or preferably monoclonal, can be intact antibody molecules or fragments containing the active binding region of the antibody, e.g. Fab or Ffab'^»°9 can be produced using methods known in the art [see e.g. RAW. DeWeger et al. "Eradication Of Murine Lymphoma And Melanoma Cells By Chlorambucil-Antibody Complexes, Immunological Rev., 62, pp. 29-45 (1982) (tumor-specific monoclonal antibodies prepared and used in conjugates), M. Yeh et al. "Cell Surface Antigens Of Human Melanoma Identified By Monoclonal Antibodies", Proe. Nati. Acad. Sei., 76, p. 2927 (1979), J.P. Brown et al. "Structural Characterization Of Human Melanoma-Associated Antigen p97 with Monoclonal Antibodies", J. Immunol., 127 (No. 2), pp. 539-546 (1981) (production of tumor-specific, monoclonal antibodies), and J.P. Mach et al. "Improvement Of Colon Carcinoma Imaging: From Polyclonal Anti-CEA Antibodies And Static Photoscanning to Monoclonal Fab Fragments And ECT", in Monoclonal Antibodies for Cancer Detection And Therapy, R.W. Baldwin et al. (ed.), pp. 53-64 (Academic Press 1985) (production of antibody fragments and use thereof for localization of tumor cells) ].If monoclonal antibodies are used, the antibodies may also be e of mouse origin or human origin, or be chimeric antibodies [see e.g.

V.T. Oi "Chimeric Antibodies", BioTechniques, 4, (No. 3), pp. 214-221 (1986)].

The enzyme component of such immunoconjugates includes any enzyme capable of acting on a prodrug in such a way as to convert it to its more active, cytotoxic form. The term "prodrug" here denotes a precursor of, or a derivative form of, a pharmaceutical active substance, which precursor or derivative form, in comparison with the parent drug, is less cytotoxic towards tumor cells and is able to be enzymatically activated or converted into the more active parent form [see e.g. D.E.V. Wilman "Prodrugs in Cancer Chemotherapy", Biochemical Society Transactions, 14, pp. 375-382 (615th Meeting, Belfast 1986) and V.J. Stella et al. "Prodrugs: A Chemical Approach To Targeted Drug Delivery", Directed Drug Delivery, R. Borchardt et al. (ed.), pp. 247-267 (Humana Press 1985)].

Enzymes useful in the delivery method include penicillin amidases such as penicillin V amidase and penicillin G amidase, which are useful for the transformation of drugs which are derivatized at their amine nitrogen atoms with, respectively, phenoxyacetyl or phenylacetyl, for over-the-counter medicines.

According to the above, a penicillin amidase enzyme was covalently bound to the monoclonal L6 antibody, and the resulting immunoconjugate was used to convert a previously unknown adriamycin prodrug produced according to the invention into the active antitumor drug. adriamycin. The particular amidase used was a penicillin V amidase ("PVA") isolated from Fusarium oxysporum which hydrolyzes phenoxyacetylamide bonds. Thus, the prodrug in particular used was N-(p-hydroxyphenoxyacetyl)-adriamycin ("APO"), which was hydrolyzed by the amidase to release the powerful anti-tumor agent adriamycin. The L6-PVA immunoconjugate showed no loss of enzymatic activity when compared with the activity of the unconjugated enzyme, and most of the binding activity of the L6 antibody was retained in the conjugate.

According to the in vitro studies performed, treatment of human lung tumor cells with the L6-PVA conjugate, followed by exposure of the cells to the APO prodrug, resulted in cytotoxicity comparable to that obtained by treating the cells with adriamycin alone. It is important to note that the APO prodrug alone exhibited far lower cytotoxicity against the tumor cells.

Corresponding in vitro studies were also carried out using an 1F5-PVA conjugate in which the PVA enzyme was conjugated with 1F5 which is a monoclonal antibody that reacts with an antigen found on lymphoma cells. Treatment of Daudi lymphoma cells with the lF5-PVA conjugate, followed by exposure of the cells to APO, resulted in cytotoxicity comparable to that obtained by treatment with adriamycin alone, while treatment of cells with APO alone resulted in very little cytotoxicity.

Although the synthesis and use of the hitherto unknown adriamycin prodrug, N-(p-hydroxyphenoxy-acetyl)-adriamycin, is described here, it is obvious that the present invention encompasses the synthesis and use of other related adriamycin prodrugs which can be derivatized in essentially the same way. The prodrug N-(phenoxyacetyl)-adriamycin, for example, also falls within the framework of the invention, as the prodrug can be synthesized using the regulation described here by replacing the reactant p-hydroxyphenoxyacetic acid (see example 4

in the following) with phenoxyacetic acid. It is also obvious that the adriamycin prodrugs according to the invention comprise other N-hydroxyphenoxyacetyl derivatives of adriamycin, e.g. derivatives substituted in different positions of the phenyl ring, as well as N-phenoxyacetyl derivatives.

The present embodiment further comprises the use of other amidases such as penicillin G amidase as an enzyme component in the immunoconjugate.' When, for example, a penicillin G amidase is used as an enzyme, the prodrug should contain a phenylacetylamide group (as opposed to the phenoxyacetylamide group in APO), as penicillin G amidase hydrolyzes this type of amide bond [see e.g. EEL. Margolin et al., Biochim. Biophys. Acta, 616, pp. 283-89 (1980)]. Other prodrugs produced according to the invention thus include N-(p-hydroxyphenylacetyl)-adriamycin, N-(phenylacetyl)-adriamycin and other optionally substituted N-phenylacetyl derivatives of adriamycin.

The invention is further described in the following example.

Example

The present example illustrates the production of

antibody-enzyme conjugates and production of prodrugs by the analog method according to the invention, as well as conversion of a relatively non-cytotoxic prodrug into an active anti-tumor agent which exhibits in vitro cytotoxicity against tumor cells. In accordance with this example, an L6-penicillin V amidase (hereinafter referred to as "PVA") immunoconjugate was used to convert an N-phenoxyacetyl derivative of adriamycin into the known anti-tumor agent adriamycin.

Preparation of antibody-penicillin V amidase conjugates

In accordance with the present example, an L6-PVA immunoconjugate and an IF5-PVA conjugate were prepared. L6 is a monoclonal antibody of the IgG2a subclass, which antibody is specific for and binds to a glycoprotein antigen on human lung carcinoma cells [see I. Hellstrom et al. (1986), supra]. 1F5 is a monoclonal IgG2a antibody specific for CD-20 antigens on normal and neoplastic B cells [see E.A. Clark et al., "Role Of The Bp35 Cell Surface Polypeptide In Human B-Cell Activation", Proe. Nati. Acad. Pollock. USA, 82, pp. 1766-70 (1985)]. The L6 hybridoma that forms the monoclonal L6 antibody has been deposited at the American Type Culture Collection (ATCC) under deposit no. HB8677 in connection with the submission of European patent application no. 207963. The lF5 hybridoma which forms the monoclonal lF5 antibody was deposited at ATCC on 12 February 1988 under deposit no. HB9645. The amidase enzyme used was a penicillin V amidase isolated from a fungal culture of Fusarium oxysporum in accordance with the procedures described by D.A. Lowe et al., "Enzymatic Hydrolysis of Penicillin V to 6-aminopenicillanic Acid by Fusarium oxysporum", Biotechnology Letters, 8 (3), pp. 151-156 (1986). Fusarium

<oxysporum strains from which this enzyme can be isolated are deposited at ATCC. PVA is thus easily available

enzyme that converts penicillin-V to pencillanic acid. Specifically, PVA hydrolyzes the phenoxyacetylamide bond in penicillin-V to form pencillanic acid. The enzyme that reacts with phenoxyacetamides can therefore be used to cleave prodrugs of known cytotoxic agents that are derivatized with phenoxyacetic acid or p-hydroxyphenoxyacetic acid.

The antibody-PVA conjugates are prepared by covalently linking the prodrug to the monoclonal antibodies L6, 96.5 or 1F5 via a thioether bond using a method similar to the method described in J.M. Lambert et al., "Purified Immunotoxins That Are Reactive With Human Lymphoid Cells", J. Biol. Chem., 260 (No. 22), pp. 12035-12041

(1985). Antibodies L6 and 1F5 were reacted with iminothiolane as described, and the number of introduced sulfhydryl groups in each antibody was determined to be between 1 and 2, using Ellman's reagent [see P.W. Riddles et al., "Ellman's Reagent: 5,5'-Dithiobis(2-nitrobenzoic Acid)-A Reexamination", Analytical Biochemistry, 94, pp. 75-81 (1979)].

A 9 mg/ml solution of the PVA enzyme was then prepared in PBS and was treated with SMCC (succimidyl-4-(N-maleimidomethyl)-cyclohexane-1-carboxylate, 100 mM in DMF) so that the final concentration was 5 mM. Treatment with SMCC introduced maleimido groups into the enzyme. After 30 minutes at 30°C, the modified enzyme was purified by gel filtration on "G-25 PD-10 Sephadex" and was eluted with PBS. The modified PVA was then added to a solution of each of the thiolated antibodies in a molar ratio of 3:1. Each reaction mixture was saturated with nitrogen and allowed to stand at room temperature for 3 hours and then incubated at 4°C for an additional 18 hours. At this point, 2-aminoethanethiol (final concentration 1 mM) was added to each solution to block any further unreacted maleimides.

Each reaction mixture was then passed through a gel filtration column ("G-25") eluting with 20 mM

Tris, pH 7.2, with 50 mM NaCl. The resulting mixtures

was purified on "DEAE Sephadex" columns (2.5 x 10 cm).

The fractions were monitored at 280 nm. The unreacted antibody in each mixture was not bound to the column, and the conjugate and unreacted PVA were eluted with 20 mM Tris, pH 7.2, with 0.5 M NaCl. The fractions containing PVA and conjugate were then concentrated using an "Amicon YM-30" ultrafiltration filter and were purified on a

"Sephacryl S-300" column (2.5 x 95 cm) eluting with PBS. The fractions were monitored at 280 nm, and the fractions containing pure conjugate as determined by SDS-PAGE (4-12% gradient gel) were combined.

Preparation of a previously unknown adriamycin prodrug

Each of the antibody-PVA conjugates prepared above was then reacted with a previously unknown adriamycin prodrug. More specifically, the prodrug used was N-(p-hydroxyphenoxyacetyl)-adriamycin (hereinafter referred to as "APO"), where the amino group in the sugar part of adriamycin is acylated with p-hydroxyphenoxyacetic acid as indicated in Figure 2.

This adriamycin prodrug was prepared as follows:

84 mg was added to 10 ml of tetrahydrofuran

(0.5 mmol) of p-hydroxyphenoxyacetic acid, 57 mg (0.5 mmol) of N-hydroxysuccinimide and 100 mg (0.49 mmol) of dicyclohexylcarbodiimide. This mixture was stirred for 2 hours, after which the solution was filtered and the filtrate was added to 200 mg (0.35 mmol) of adriamycin hydrochloride. 0.1 ml of triethylamine was added to the reaction mixture, and stirring was continued for 4 hours. The reaction mixture was then filtered through glass wool and evaporated to a residue under high vacuum. The resulting mixture was purified on a "Silicagel 60" column (2.5 x 20 mL) eluting with 95:5 dichloromethane:methanol. The combined fractions were again purified on the same type of column, whereby 70 mg (0.1 mmol, 30% yield) of pure N-(p-hydroxyphenoxy-acetyl)-adriamycin was obtained.

FAB MS m/e 694.2125 (M+H)<+>. Calculated for <C>35<H>36N014' 694'2136- 360 MHz 1h nmr (CDCIj) S : 1.06 (d, 3H, sugar CR"3) , 1.5-2.2 (m, 6H, sugar H) , 3.0 (q, 2H) , 4.0 (s, 3H, 0CH3), 4.35 (s, 2H, COCH2O), 4.8-5.0 (m, 3H), 5, 2 and 5.4 (s, 1H), 6.6-6.8 (dd,;4H, phenoxy Arh), 7.4-7.9 (m, 3H, 2,3,4-H), 9 .0 (s, 1H, Ar<1>OH), 11.61 and 12.39 (s, 1H, ArOH).

It will be obvious that other phenoxyacetylamide derivatives of adriamycin can be prepared by using substantially the same method as described above. For example, N-(phenoxyacetyl)-adriamycin can be prepared as described above, by replacing p-hydroxyphenoxyacetic acid with 0.5 mmol (76 mg) of phenoxyacetic acid.

Reaction of an antibody penicillin V amidasecon conjugate with an adriamycin prodrug

The ability of the antibody-PVA conjugate, L6-PVA, to convert the previously unknown prodrug APO to adriamycin was determined as follows: either a) PVA alone (final concentration: 50 ug/ml), b) 100 Mg/ml was added L6-PVA conjugate (final PVA concentration: 25 yg/ml) or

c) 10 µg/ml L6-PVA (final PVA concentration: 2.5 µg/ml) to a solution of APO (0.1 mM) in PBS. Each reaction was monitored by HPLC using a "Phenominex C-18" column (3 µm, 4.5 x 100 mm) and gradient elution with 20-60% tetrahydrofuran in water with 0.1% H 3 PO 4 (1, 0 ml/minute, detection at 495 nm). Under these conditions, the adriamycin was eluted after 8.9 minutes, and the APO was eluted after 12.2 minutes. The results appear from figure 3: As can be seen from the figure, the amide group in APO was hydrolysed with the help of PV, which is evident from the formation of adriamycin. Under the conditions used, the half-life for hydrolysis of APO by means of PVA was approx. 20 minutes. Furthermore, it was found that within 40 minutes from the start of the reaction, both the enzyme alone and the antibody PVA conjugate could cause hydrolysis of at least 80% of APO to adriamycin. 10 ug/ml (2.5 pg/ml PVA) could effect hydrolysis of this magnitude within 120 minutes. Finally, it is clear from these investigations that antibody

The PVA conjugate according to the invention showed no apparent loss of enzymatic activity as a result of the binding of the enzyme to the antibody, which is evident from the fact that the conjugate and the free enzyme showed similar properties with regard to hydrolyzing APO to adriamycin.

Serum stability of the hitherto unknown adriamycin prodrug

The stability of APO in human serum was determined using HPLC and measuring the rate of APO disappearance and the rate of formation of adriamycin. A solution of APO (10 mM in dimethylformamide) was thus added to fresh human serum so that the final concentration was

0.1 mM. Aliquots (50 µl) were diluted with methanol (50 µl) to precipitate serum proteins. These samples were then centrifuged and analyzed by HPLC as described above. No hydrolysis of APO to adriamycin occurred within 2 hours.

Binding of the antibody-PVA conjugates to H2981 tumor cells

The ability of the L6-PVA and 1F5-PVA conjugates of the invention to bind to H2981 tumor cells was measured

as follows: The immunoconjugates or free antibodies were serially diluted in modified Dulbecco's incomplete medium (IMDM), and 100 µl aliquots were incubated at 4°C with 10<6 >30 minutes. The cells were washed and incubated with 50 µl FITC goat anti-mouse antibody (diluted 1:12.5) for an additional 30 min at 4°C. The cells were washed and analyzed on a Coulter Epics-C cell fluorescence analyzer. Dead cells were excluded, and the logarithmic mean value of the green fluorescence intensity of each sample was obtained. This number was converted to a linear scale, and ratios between the negative control (cells + FITC goat anti-mouse antibody) and all test samples were calculated. The results of the bond test are depicted in figure 4.

FACS analysis indicates that both the L6 and L6-PVA conjugate showed strong binding to the tumor cells, while the lF5-PVA conjugate did not show binding of any significant magnitude. This binding study indicates, firstly, that the conjugation to the enzyme did not in any significant way affect the binding ability of the antibody component 1 immunoconjugates. Second, this sample again demonstrates the specificity of the binding of the conjugates, in that the L6-PVA conjugate binds to the L6-positive H2981 tumor cells, and the lF5-PVA conjugate, due to the lack of specificity of the lF5 antibody for the tumor cells, essentially does not exhibits some bonding.

In Vitro Cytotoxicity of the Antibody-PVA Conjugate/Adriamvicin-Prodrug Combination on H2981 Tumor Cells

The in vitro cytotoxicity of the antibody-PVA/adriamycin prodrug combination against H2981 tumor cells was measured using a <3>H-thymidine uptake assay. Briefly, H2981 tumor cells were plated on 96-well microtiter plates in IMDM (10,000 cells/well) and allowed to attach for 18 hours at 37°C. The antibody-PVA conjugates, L6-PVA or 1F5-PVA,

was then added at a concentration of 10 µg/ml antibody, and the plates were incubated for 30 minutes at 4°C.

The wells were then washed 4 times with IMDM, and APO was added at various concentrations in IMDM. After 2 hours, the wells were again washed, IMDM was added, and the cells were allowed to stand for 18 hours at 37°C. At this point, H-thymidine (1 pCi/well) was added, and after 6 hours the plates were frozen at -70°C to detach the cells. After thawing, the cells were harvested on glass fiber filters. Incorporation of ^H-thymidine was measured using a "Beckman 3801" scintillation counter and was compared to cells treated with APO or adriamycin (ADM) alone.

Using this assay, the inhibition of <3>H-thymidine incorporation into tumor cell DNA was measured, and the cytotoxic effect of the prodrug APO on cells with or without pretreatment of the cells with the L6-PVA or lF5-PVA conjugates was then measured. . The cytotoxic effects of these combinations were compared with the cytotoxicity observed when treating the cells with the parent drug adriamycin alone. As can be seen from Figure 5, adriamycin with an IC^Q value of 38 nM was signi-

significantly more toxic towards tumor cells that had not been treated with conjugate, than APO with an IC50~ value of 2 pM. This was expected in view of previous reports, from which it appears that adriamycin amides are less toxic than adriamycin (see, for example, Y. Levin and B.A. Sela, FEBS Letters, 98, p. 119 (1979) and R...Baurain et al., J. Med. Chem., 23, p. 1171 (1980)). Pretreatment of the cells with L6-PVA increased the cytotoxicity of APO 20-fold, to a level comparable to the cytotoxicity of adriamycin alone. Pretreatment of cells with 1F5-PVA did not affect the toxicity of APO in any way. These results indicate that the L6-PVA conjugate will be able to hydrolyze the relatively non-cytotoxic prodrug APO for the destruction of tumor cells to an extent comparable to the use of adriamycin alone, and that this cytotoxicity is antigen-specific, as evidenced by the fact that lF5-PVA- the conjugate that did not significantly bind to this particular tumor cell line did not cause such cytotoxicity.

Binding of the antibody-PVA conjugates to Daudi lymphoma cells

The ability of the L6-PVA and 1F5-PVA conjugates to bind to the known Daudi cell line was also measured. This cell line is a Burkitt's lymphoma cell line deposited at ATCC (ATCC No. CCL 213) and which expresses the CD-20 antigen to which the 1F5 antibody binds. The results are depicted in Figure 6.

In this case, both the 1F5 monoclonal antibody and 1F5-PVA conjugate bound strongly to the lymphora cells. This investigation again indicated that the binding ability of the conjugates was not significantly affected by the conjugation method.

As can be seen from the figure, the L6 antibody exhibited

and the L6-PVA conjugate additionally showed no apparent binding to the Daudi cells. This could also be expected as Daudi tumor cells do not display the antigen with which the L6 antibody reacts. This study, combined with the binding studies described above, clearly demonstrates the binding specificity of the conjugates, i.e. that L6-containing conjugates bind specifically to L-positive tumor cells, and lF5-containing conjugates bind specifically to CD-20-positive tumor cells.

In vitro cytotoxicity of the antibody-PVA conjugate/adriamycin-proleq agent combination on Daudi cells

The in vitro cytotoxic effect of the L6-PVA or lF5-PVA conjugate in combination with the APO prodrug was then tested on Daudi cells.

The <3>H-thymidine assay was performed essentially as described above, with slight modifications due to the fact that the Daudi cells are non-adherent. Thus, approx. 250,000 Daudi cells in IMDM were placed in each well of a 96-well microtiter plate, and the antibody-enzyme conjugate was added. The reaction mixture was incubated at 4°C for 30 minutes. Unbound anti-substance f-enzyme conjugate was removed by centrifugation at 500 x g for 5 min, and the supernatant was removed. The cells were re-suspended in IMDM and the washing procedure was repeated 3 times to remove all unbound conjugate. Then, APO in IMDM was added, and after resting for 2 hours, the cells were washed once as described above. The resulting part of the sample was performed as described previously.

Using this test, the inhibition of <3>H-thymidine incorporation into Daudi cells' DNA was measured, and thus the cytotoxic effect of the APO prodrug on the cells with or without pretreatment of the cells with L6-PVA- or lF5-PVA- the conjugate. The cytotoxic effects of these combinations were compared with the observed cytotoxicity after treatment of the cells with adriamycin alone. As can be seen from figure 7, adriamycin was significantly more toxic than APO towards Daudi cells that had not been treated with any conjugate. Pretreatment of the cells with the lF5-PVA conjugate significantly increased the cytotoxicity of the APO prodrug to a level comparable to the cytotoxicity of adriamycin alone, whereas pretreatment with the L6-PVA conjugate did not result in such an increase.

It should be noted that the results obtained in these binding and cytotoxicity studies are the opposite of the results obtained with these conjugates in the previously described studies using H2981 tumor cells where the L6-PVA plus APO combination showed increased cytotoxicity and 1F5- The PVA plus APO combination did not show such an effect. This could be expected given the different specificities of the L6 and 1F5 antibodies in the conjugates and clearly demonstrates the specificity of the cytotoxic effects achieved with the conjugate/prodrug combinations.

This study further suggests the utility of the 1F5-PVA conjugate in combination with APO in producing cytotoxic effects on tumor cells in vitro. Thus, the in vitro cytotoxicity studies in this example demonstrate that any of the conjugates obtained according to the invention containing an antibody that reacts with a tumor-associated antigen can be used for the treatment of tumors with which this antibody reacts.

Claims (3)

1. Analogous process for the preparation of a therapeutically active anthracycline derivative of the formula wherein R is the group R<1> is H, and R<3> is OH or OCH3, or R<1> is OH, and R<3> is OCH3, and R <2> is H or OH. characterized in that a compound of the formula in which R<1> and R3 have the above meaning, or an acid addition salt thereof, is reacted with a compound of the formula in which R<2> has the above meaning, or an acylating equivalent thereof. 2. Process according to claim 1 for the production of N-(p-hydroxyphenoxyacetyl)-adriamycin, characterized in that corresponding starting materials are used. 3. Process according to claim 1 for the production of N-(phenoxyacetyl)-adriamycin, characterized in that corresponding starting materials are used.