292
`
`--293
`
`which is unsatisfactory for pharmacokinetic reasons [4]. Systematic studies on the
`mechanism revealed that it involved an initial hydration of the 5,6—imine double
`
`bond, followed by ring opening to the formyl derivative [5]. By adding bisulphite
`across the double bond protonated at pH 2.5, the derivative is stabilised and rapidly
`reverts to azacytidine under physiological conditions (Fig. 1). Not only is this
`azacytidine prodrug stable, and therefore suitable for prolonged infusions, but be-
`ing more soluble requires a much smaller volume of perfusing solution. Further-
`more, because of its stability at acid pH it might prove to be a stable prodrug for
`oral administration [5].
`
`Similar techniques have been used to make soluble prodrugs of anticancer agents
`whose low water solubility limits their clinica.l.._,usefulness. Thus, adenosine
`arabinoside has been studied both as an antiviral and an antitumour agent. It is
`
`employed most effectively by intravenous infusion, but its low solubility (0.5
`mg/ml) and high therapeutic dose (2.5 g) has necessitated the infusion of several
`litres of solvent [6]. The 5 ’ —0—formate ester is some 60 times more soluble and, since
`it is converted rapidly to the parent drug in vivo and is not subject to enzymatic
`deamination, it is clearly a suitable prodrug (Fig. 2).
`NH
`
`NH2
`
`2
`
`/
`
`N
`
`\
`
` >
`
`/
`
`N
`
`\
`
`LN i N>
`
`-—————-5
`
`HOCH2
`
`0
`
`OH
`
`OH
`
`0
`
`()H
`
`OH
`
`0l
`
`l
`HCOCH2
`
`Fig. 2
`
`Some antitumour agents are of interest using experimental models but are not us-
`ed clinically because of their low solubility. Acronycin, for example,
`is active in
`12/ 17 animal tumour systems [7}, but because of its low solubility (2— 3 mg/ litre)
`has only been evaluated clinically by oral administration in capsules. Because it is
`likely that absorption from the gut will be poor and variable between patients,
`clinical trials have been difficult to evaluate. In order to overcome this problem pro~
`
`drugs suitable for intravenous administration have been developed. Of particular in-
`terest have been the acetylacronium derivatives, which have molar solubilities
`several orders of magnitude greater than acronycin and which under physiological
`conditions rapidly revert to the parent compound [8}.
`Water-soluble prodrugs have also been made to overcome problems of toxicity
`associated with the parent compound. The electron affinic nitroimidazole radiation
`sensitisers often cause peripheral neurotoxicity as an unwanted side effect. More
`
`Petitioner Mylan Pharmaceuticals Inc. — Exhibit 1012 — Page 152
`
`phasise prodrugs which improve the pharmacokinetic properties of anticancer
`agents and, in particular, prodrugs which are activated selectively in tumour cells
`to the active drug.
`
`2. Prodrugs with altered chemical stability, solubility or better tissue penetration
`
`The chemical instability of some anticancer agents when dissolved in physiological
`saline is well recognised and is overcome, for example,
`in the case of reactive
`alkylating agents by ensuring rapid dissolution and injection, and, in the case of for-
`mulations of dimethyltriazenoimidazole carboxamide that are light unstable, by the
`use of dark vessels and the avoidance of sunlight. Other agents may decompose less
`slowly, but nevertheless this may be a problem when they are used as infusions over
`long periods of time. This problem can usually be overcome by a study of the
`mechanism of decomposition, followed by appropriate prodrug design to increase
`stability. Azacytidine, for example,
`is used in the treatment of acute myeloid
`leukaemia. When given by an intravenous bolus it causes severe and dose—limiting
`gastrointestinal toxicity. However, if the drug is slowly infused over a 5-day period
`this toxicity is eliminated, but such infusions are not really practicable because of
`the instability of azacytidine in aqueous solutions. Azacytidine is hydrolysed rever-
`sibly to the ring-opened formyl derivative, which is then converted slowly and ir-
`reversibly to the guanylurea (Fig. 1). While both the formyl and the urea derivatives
`have lower toxicity, they do not have antitumour activity so that over the long infu-
`sion period a gradually reducing concentration of azacytidine is being infused,
`
`NH2
`
`Petitioner Mylan Pharmaceuticals Inc. - Exhibit 1012 - Page 152
`
`

`
`
`
`295
`
`/
`
`HGPRT
`
`SH
`
`N)\/l[N\>
`|‘\N
`ll‘!Sug. Phosphate
`
`SH
`
`N/I N\>
`KN
`N
`
`Fig. 5
`
`mechanisms. However, some tumours, while initially very responsive to this agent,
`become resistant due to the outgrowth of cells which have lost the activating
`
`HGPRT enzyme. This resistance cannot be overcome by using the nucleotide of
`6—mercaptopurine since the ionised form does not penetrate cells and is quickly
`degraded by extracellular enzymes. However,
`the 2’-0—acy1—6—thioinosine cyclic
`3 ’ ,5 ’ -phosphate is a very effective prodrug against cells with acquired resistance to
`6—mercaptopurine [11]. The lipophilic prodruct is readily taken up by the resistant
`cells and is then converted by phosphodiesterase to 2’-0-acylthioinosine and by
`acylases to thioinosinic acid (Fig. 6).
`
`SH
`
`0
`
`OH OR
`
`0
`ll
`HOFFOCH2
`OH
`
`5
`
`H
`
`N
`
`-———>
`
`o
`H
`HOPOCHZ 0
`OH
`
`OH OH
`
`SH
`
`0
`
`/CH2
`0\
`
`O = P--O OR
`I
`OH
`
`Fig. 6
`
`In the above example the addition of the 2’—0-acyl group improves drug penetra-
`tion. Some anticancer agents are hydrophilic and concentrate in cells by an active
`transport mechanism [12]. Resistance may occur by the loss of this mechanism and
`might possibly be overcome by the design of a liphophilic prodrug. Melphalan (L-
`phenylalanine mustard), for example, is taken up by an active transport mechanism
`[13], and if resistance arose by loss of the transport mechanism N-acyl esters of
`melphalan might possibly be active against resistant cells.
`
`Petitioner Mylan Pharmaceuticals Inc. — Exhibit 1012 — Page 153
`
`
`
`SCH2O0CC(CH3)3
`N
`
`“(I
`\N
`
`til
`
`SCHZOH
`N
`
`SH
`
`\> —-~»'d\>-—~O\>
`\N
`iii
`\N
`N‘
`
`CH200CC(CH3i3
`
`CHZOH
`
`Fig. 4
`
`
`
`294
`
`
`
`
`
`hydrophilic nitroimidazoles concentrate less in neural tissues and are excreted more
`rapidly, both properties reducing neurotoxicity. These nitroimidazoles may also be
`suitable for oral administration, but since decreased lipophilicity will decrease oral
`absorption a prodrug may be required to obtain optimal activity [9]. A
`nitroimidazole has been synthesised which has a very low octanol/water partition
`coefficient and might be expected to be less neurotoxic than earlier radiation sen-
`sitisers such as misonidazole, which has dose—limiting neurotoxicity. In order to in-
`crease absorption from the gut the prodrug acetyl ester was synthesised and shown
`to be converted completely to the drug in the first pass through the liver (Fig. 3).
`In some cases it is necessary to reduce the water solubility of a drug by the syn-
`thesis of a more fat—soluble prodrug. Anticancer agents such as 6—mercaptopurine
`are also useful in the treatment of psoriasis, but, because they do not penetrate the
`skin, must be given systemically, with_resulting toxicity. A soft alkylated derivative
`of 6—mercaptopurine may be used as a topical treatment of psoriasis [10] since it is
`transported effectively through epidermal barriers and is then converted, firstly by
`esterases and then spontaneously to the active thiopurine (Fig. 4).
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`3. Prodrugs which overcome acquired resistance
`
`Often if the mechanism by which a cell acquires resistance to a drug is understood
`it is possible to design a prodrug to overcome this resistance. Thus, 6—mercapto-
`purine, like many antipurines and antipyrimidines, requires intracellular conversion
`to a nucleotide before it exerts its inhibitory action. 6—Mercaptopurine is converted
`by hypoxanthine, guanine phosphoribosyl transferase (HGPRT) to the nucleoside
`monophosphate (Fig. 5), which interferes with cell growth by a number of
`
`
`
`N
`
`N
`
`__._L£9r_____,
`
`</
`
`\>\
`
`N0
`
`2
`
`N
`I
`CH2CONHCH2CH20AC
`
`Fig. 3
`
`</
`
`\>\
`
`N0
`
`2
`
`N
`I
`CHZCONHCHZCHZOH
`
`Petitioner Mylan Pharmaceuticals Inc. - Exhibit 1012 - Page 153
`
`

`
`296
`
`4. Prodrugs with improved pharmacokinetic properties
`
`Most anticancer drugs are cell cycle-specific, meaning that they are particularly toxic
`to cells in cycle. Some drugs are even more specific, being cytotoxic to cells only in
`a particular phase of the cell cycle. For these agents the correct pharmacokinetic
`profile is essential if selective inhibition of tumour cells is to be obtained. Cytosine
`arabinoside (Ara-C), for example, acts after its intracellular conversion to the
`triphosphate, in which form it is an inhibitor of DNA polymerase. Therefore, it is
`S phase-specific and to achieve its optimal anti-tumour effect it must be present in
`the tumour environment at a toxic concentration for the period of time that it takes
`all the cells of the asynchronously growing tumour to enter and pass through the
`S phase. However, Ara—C as well as being activated to its triphosphate is also a
`substrate for cytidine deaminase and is converted by this enzyme to an inactive
`product (Fig. 7). This would result in a rapid clearance of Ara-C from the body
`and its profile after intraperitoneal or intravenous injection would be as indicated
`diagrammatically in Figure 8. After either route of administration the phar-
`
`Concentration
`
`\
`\
`\
`‘§-—i.V.
`
`injection
`
`I. P.
`
`iniection
`
`4"-‘ Optimal concentration
`
`
`
`297
`
`macokinetic profile is not satisfactory, since there will be periods of time between
`injections when a tumour-inhibitory concentration of the drug will not be present.
`Any cells passing through the S phase during this time will escape the cytotoxic ef-
`fects of the drug. Therefore, the optimum dose schedule of Ara—C is by slow infu-
`sion over a period of many hours. However, long periods of infusion require the
`patient to stay in hospital, and extensive efforts have been made to discover pro-
`drugs of Ara—C and other phase-specific agents which will have more favourable
`pharmacokinetic properties.
`
`An early attempt was the synthesis of the 2,2’—anhydro analogue of Ara—C (Fig.
`9), which is resistant to cytidine deaminase and slowly hydrolyses to release the ac-
`tive drug [14]. Other prodrugs of Ara—C includeea large number of derivatives
`substituted or disubstituted in 5’, 3’ and 2’ of the arabinose or the N4 of the
`cytidine ring (Fig. 9) [15 — 19]. Amongst this large number of chemicals the most
`lipophilic, such as the 5’—adamantoyl derivative, proved to be the most effective in
`increasing the therapeutic index against tumour—bearing animals [16]. A relationship
`has been shown between low water solubility and high antitumour activity [20].
`However, this poor water solubility may pose a problem in the clinical trials of these
`prodrugs.
`
`NH2
`
`”/ I
`OAN
`
`NH
`
`“U
`|\N
`
`NHOCR
`
`“’ I
`0J\N
`
`ROCHZ
`
`0
`
`R0
`
`OR
`
`Fig. 9
`
`HOCHZ 0
`
`HOCHZ 0
`
`0
`
`OH
`
`H0
`
`OH
`
`like Ara—C, degraded rapidly by adenosine
`Adenosine arabinoside (Ara-A) is,
`deaminase to an inactive metabolite (Fig. 10). Using the same approach as for Ara-
`C, many analogues have been synthesised as potential ‘slow release’ forms of Ara-A
`and include 2’, 3’ and 5’ mono- or disubstituted O—acyl derivatives [21, 22]. Of par-
`ticular interest is the 5’—O—valeryl analogue, which shows a marked increase in
`potency compared to Ara-A. Not only is the drug acting as a slow release form of
`Ara-A (Fig. 10), but the prodrug inhibits the degradation of Ara-A to hypoxanthine
`arabinoside. Thus, as long as the prodrug is present there will be no degradation
`of Ara-A and there will be a corresponding increase in potency [22, 23].
`A series of 5’ (steroid 2’—phosphoryl) derivatives of Ara—C have been synthesised
`
`Petitioner Mylan Pharmaceuticals Inc. — Exhibit 1012 — Page 154
`
`Petitioner Mylan Pharmaceuticals Inc. - Exhibit 1012 - Page 154
`
`

`
`299
`
`ferences in distribution compared to daunorubicin [34].
`The desire to alter the distribution of 6—mercaptopurine and hopefully to localise
`it selectively in lymphatic tissues and the central nervous system, which may be the
`site of secondary tumours, has been the aim behind the synthesis of a series of
`dinitrobenzoate esters of varying chain length [35].
`
`5. Selective activation in cancer cells
`
`Many prodrugs have been synthesised which are themselves of low toxicity, but
`which may be converted enzymatically in tumour-c~ells to an active drug. One of the
`first examples of this approach was the synthesis of diethylstilbestrol diphosphate,
`which releases active diethylstilbestrol in prostatic cancers high in acid phosphatase
`
`[36] (Fig. 11).
`
`_
`
`H
`.
`$2 5
`f2“5
`NaOP02O c = c OPO20Na
`
`Phosphatase
`
`C2H5
`i
`
`CZH5
`l
`
`HO -0 c = c
`
`OH
`
`Fig. 11
`
`There is no doubt that the presence or absence of activating enzymes can greatly
`alter the response of cancer cells to a prodrug. Thus, 5’-deoxy-5—fluorouridine
`(5-FUdR) is a good antitumour prodrug because it is cleaved selectively in most
`tumours by uridine phosphorylase to 5—fluorouracil (5-FU) which, after conversion
`to its nucleotides, is a cytotoxic antimetabolite (Fig. 12). However, whereas human
`B lymphocytes are sensitive to both 5-FU and 5-FUdR, the L1210 leukaemic cells
`do not respond very well to the latter because they do not convert it to 5-FU and
`its nucleotides
`[37].
`In fact,
`5—FUdR is not a good substrate for uridine
`phosphorylase, and derivatives such as the 5’—0-tosyl are more active on cells such
`as the L1210 which respond only poorly to 5—FUdR [38].
`The principles employed in the design of prodrugs activated in tumour cells are
`best illustrated using the alkylating nitrogen mustards as examples. This class of an-
`titumour agent acts by covalent binding to cellular macromolecules and causes
`cytotoxicity principally because they cross—link DNA [39]. If these agents are not
`metabolised in vivo a broad correlation can be seen between their chemical reactivity
`
`Petitioner Mylan Pharmaceuticals Inc — Exhibit 1012 — Page 155
`
`
`
`
`
`/
`
`1/
`
`,
`
`NH2
`
`N
`
`/
`
`no /
`
`KN
`
`N
`
`"
`4-.»
`
`CH3(CH2)3C0 0 CH2
`
`0
`
`H0
`
`OH
`
`Fig. 10
`
`H0 CH2
`
`0
`
`HO
`
`OH
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`from steroids such as cortisone, prednisolone and a range of anti—inflammatory
`steroids [24]. Some but not all were superior to Ara—C alone in tumour—bearing
`animals, and it is of interest that prodrug activation appeared to take place inside
`the cells rather than in the blood. Clearly, this could form the basis for the design
`of prodrugs with activity against cells, with acquired resistance to Ara—C as a result
`of loss of the activating enzyme. Some prodrugs of Ara-C, such as Ara—CTP-L-
`dipalmitine, are not substrates for cytidine deaminase and are very effective against
`tumour cells in culture [25]. They might be readily taken up by cells and then
`
`degraded to Ara-CTP. In such a situation cells resistant to Ara—C because of loss
`of the kinase would remain sensitive to this type of prodrug. In fact, diacylglycerol
`derivatives of cystosine arabinoside do have an effect against some mouse tumours
`which have acquired resistance to Ara-C [26]. Other analogues of Ara-C which have
`been studied include phospholipids and polyglutamates and an N4—palmitoyl
`derivative which was effective when given orally to tumour—bearing mice [25 — 30].
`Following similar lines esters, glutamyl conjugates, amides, hydrazides and
`hydroxamic derivatives have been made of methotrexate [31 -33]. Although in
`many cases the rationale has been to produce slow~release prodrugs forms of
`methotrexate, others have been made which may be taken up readily by cells by
`
`passive diffusion, and thus be active in cells resistant to methotrexate in which the
`active transport mechanism has been lost. In the course of these studies some in-
`teresting findings have been made. Most importantly, diesters of methrotrexate may
`be cytotoxic to leukaemia cells in their own right, an effect which is only partially
`reversed by folinic acid,
`indicating that they may be acting by more than one
`pathway. Furthermore, in mouse serum there is a regioselective hydrolysis of the
`diester, the or-carboxyl ester being hydrolysed much more readily than the 'y [31].
`Studies on esterases as prodrugs have emphasised that the rodent may not be a
`satisfactory model for humans since serum esterases in the former are much more
`active than in humans.
`
`N, N—Dibenzyldaunorubicin has no activity against tumour cells in vitro but is ac-
`tive in vivo. suggesting that in whole animals it acts as a prodrug, with possible dif-
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`NH2
`
`N
`
`~t\>
`
`KN
`
`N
`
`\
`
`‘\
`
`\\
`
`OH
`
`N
`
`N/
`
`\$1»
`
`‘
`——>
`
`H0 CH
`
`2
`
`0
`
`H0
`
`0
`
`H
`
`
`
`Petitioner Mylan Pharmaceuticals Inc. - Exhibit 1012 - Page 155
`
`

`
`
`
`OH
`
`OH
`
`Fig. 12
`
`0H H
`
`
`
`
`
`
`
`
`
`
`
`and cytotoxicity. Highly reactive nitrogen mustards are likely to be very toxic (both
`to tumour cells and to normal tissues) and poorly reactive ones very much less toxic.
`Figure 13 shows that as the alkylating activity varies over a 25-fold range, the LD50
`varies from 40 to 3000 ,umol/ kg. Very small changes in structure, eg., p-
`hydroxylation, can greatly alter reactivity, and hence toxicity [40]. On this basis pro-
`drugs can be designed which are poor alkylating agents and non-toxic, but which
`act as substrates for enzymes which transform them to highly reactive and cytotoxic
`agents. Clearly, for there to be selective tumour cell kill, the prodrug must be a good
`substrate for the enzyme under physiological conditions and the enzyme ideally
`should be present uniquely in the tumour cells, or at least at much higher levels in
`the tumour than in normal tissues, especially those which are also sensitive to this
`type of agent. A further prerequisite is that once formed from the prodrug, the drug
`
`
`
`
`
`
`
`
`
`
`
`
`Compound
`
`or
`\
`CH:CH:Ci
`
`CH2CH2Cl
`/
`HOOC N\
`
`CH2CH2Cl
`
`we
`
`915
`
`4-6
`
`
`
`
`
`
`
`/CH2CH2Cl
`N\
`CH2CH2Cl
`
`/CH2CH2Cl
`
`HO N
`\CHzCHzCl
`CHzCH2BT
`
` /
`\CH2CH2Bf
`
`l3~0
`
`48-6
`
`l23 0
`
`367
`
`74
`
`39
`
`Alkylating
`activity
`(Kan X 10’)
`
`Toxicity
`(LD,.,, pmole/kg)
`
`CH3
`
`,
`
`301
`
`should react immediately and be unable to diffuse from the tumour and reach
`alkylating agent—sensitive tissues. This could be achieved by ensuring, for example,
`that the prodrug releases a drug which is unstable or is highly reactive, so that it
`would be in an inactive form if it diffused from the tumour, or perhaps charged so
`
`that it is trapped inside the cell. The approach may be illustrated by the design of
`prodrugs activated by azo—reductase.
`Following an observation that hepatocellular carcinoma in humans had azo+
`reductase levels almost as high as normal hepatocytes, a series of azobenzenes were
`
`designed to be activated by this enzyme. The plan was to design an inert prodrug
`which was a good substrate for the enzyme and which would form an alkylating
`agent so reactive that any drug diffusing from thevietumour cell would hydrolyse or
`react in the blood stream before reaching normal tissues sensitive to these agents [41,
`
`.
`
`Compound
`
`Reduction by
`Azoreductase
`tnmole lg/15')
`
`LD 50
`( mq/kq)
`
` ~ N=N M 1350
`
`COOH
`
`Q N=N M 2383
`
`COOH
`
`CH3
`
`M=-N(CH2CH2Br)2
`
`Fig. 14
`
`57
`
`13
`
`Petitioner Mylan PharmMK
`
`
`
`aceuticals Inc. — Exhibit 1012 — Page 156
`
`.
`
`..,.\
`
`Petitioner Mylan Pharmaceuticals Inc. - Exhibit 1012 - Page 156
`
`

`
`
`
`303
`
`mustard prodrug has been synthesised (Fig. 16) which when given orally will be ab-
`sorbed from the gut and pass into the liver.
`In both normal and malignant
`hepatocytes the prodrug will be activated. Normal hepatocytes are relatively resis-
`tant to nitrogen mustards and will survive doses which are toxic to the malignant
`hepatocytes. Any active drug diffusing from the normal or malignant hepatocyte
`will hydrolyse in blood to the innocuous di—2-hydroxy product before reaching other
`organs.
`
`RN=N
`
`N(cH2cHari2——> HZN
`CH3
`
`NicH2cHanz———> HZN
`CH3
`
`NicH2cHoHi2
`CH3
`
`cH3
`
`Non Toxic
`
`Fig. 16
`
`CH3
`
`Toxic
`
`CH3
`
`Non Toxic
`
`A related approach which also relies on the reduction of an alkylating prodrug
`to a reactive agent specifically in tumour cells has been described by Tsou and Su
`[43]. Based on the observation that Ehrlich ascites cells were very efficient in reduc-
`ing tetrazolium salts to formazans,
`they prepared a series of nitrogen mustards
`“which as tetrazolium derivatives would be quite inactive because of the conjugated
`
`ring system. On reduction to the formazan, the reactivity of the alkylating moieties
`would be increased and further reduction to the amine would result in a highly reac-
`
`tive and very toxic compound (Fig. 17).
`
`5.1. ACTIVATION BY HYPOXIC TUMOUR CELLS
`
`The low oxygen tension of tumours and their ability to reduce certain substrates
`more effectively than normal tissues has been observed for many yearsand is prob-
`
`/ N©
`Q< i
`N N(CH:CH:Cl):
`
`Petitioner Mylan Pharmaceuticals Inc. — Exhibit 1012 — Page 157
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`CH3
`
`Fig. 15
`
`302
`
`42]. Azobenzene mustards of the structure shown in Figure 14 will be poor
`alkylating agents in vitro because of the electron-withdrawing effects of the con-
`jugated ring system. As expected, those which are not metabolised by azo-reductase
`
`have low toxicity in vivo, as measured by their LD50 values. Some azobenzene
`mustards are good substrates for azo—reductases and in vivo are converted to toxic
`
`amines which have low LD50 values (Fig. 14). It is also necessary that the amine
`formed on reduction should be extremely reactive, having a half-life of only a few
`seconds. Azobenzenes can be synthesised which are metabolised in vivo and which
`
`can produce reactive alkylating agents of the type shown in Figure 15. The
`
`di-2-chloroethylamines have half—lives in water of between 5 and 8 minutes at pH
`7.0 and 37°C. Compared with the blood circulation time, this is quite long and
`would allow the active drug to leave the tumour cell and perhaps be taken up by
`normal tissues. The reactivity of nitrogen mustards can be increased by replacing
`the chlorine atoms by bromine (or iodine) and also by inserting a methyl group on
`the carbon atom carrying the halogen. Both these modifications lead to enhanced
`
`alkylating activity, and the 2—bromopropy1 derivative is an ideal drug to be released
`from a prodrug since its half-life is only around 40 seconds. Thus, in order to exploit
`the finding that human liver cancer has a high level of azo-reductase, an azobenzene
`
`Compound
`
`Tie in H20
`V (37°, pH7.0)
`
`NiCH2CH2CI)2
`
`CH3
`
`2
`N(CH2CH CH2
`
`8.5
`
`4.8
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Petitioner Mylan Pharmaceuticals Inc. - Exhibit 1012 - Page 157
`
`

`
`
`
`305
`
`
`the cloroethylamino group. As a result it is 30 — 40 times less toxic than HN2. There
`
`is already some indication that it may be a useful prodrug since it is active against
`
`well—established solid tumours in rats (which may be expected to have large numbers
`
` O
`
`0
`
`OH
`
`: ‘CHR
`
`0
`
`)'(
`
`——————->
`
`: ‘?HR
`
`X
`
`OH
`
`
`
`t
`CH3N(CH2CH2Cl)2
`
`————-————>
`
`CH3N(CH2CH2C|2)
`
`
`
`
`
`mcnzcnzcnz —?..> N(CH2CH2C|)2
`
`(ClCH2CH2l2N
`(C|CH2CH)2N
`
`O2N N(CH2CH2Cl)Z ——-———-D» H2N N(CH2CH2C|)2
`
`Fig. 18
`
`
`
`304
`
`
`
` ably explained by the presence in most tumours of zones of hypoxic cells. When
`
`animals bearing solid tumours are injected with lissamine green the charged dye does
`not penetrate cells but maps out well—vascularised areas of tissue. Using this tech-
`nique it has been shown that solid tumours, even when quite small, may develop
`ischaemic areas containing mainly necrotic tissues [44]. However, a layer of cells
`between the dyed and the necrotic tissue has been shown to be viable, even though
`the cells in this layer are some distance from a blood supply and consequently
`
`of hypoxic cells) and has a higher therapeutic index than HN2, suggesting selective
`reduction in the tumour. The fact that it has little activity against cells in culture
`also indicates that it is activated only in cells of low oxygen tension. Other nitrogen
`
`mustards that might be activated by a similar mechanism include the quinone, which
`could form the very much more toxic hydroquinone (Fig. 18), and possibly the
`nitrobenzene (or nitrosobenzene) mustard, which, if selectively reduced in anoxic
`tumour cells, would form the highly toxic p—phenylenediamine mustard (Fig. 18).
`One drawback with this approach is that the anaerobic bacteria of the large bowel
`are also very effective in carrying out these reductions, and thus might contribute
`to the systemic toxicity of the prodrug and reduce selectivity.
`Certain anthraquinones and mitomycin C probeably also owe some of their an-
`titumour selectivity to their reduction. Mitomycin C, for example, probably kills
`tumour cells as a result of DNA cross—linking, but in vitro this reaction can only oc-
`cur in the presence of a reducing agent. The concept of bioreductive prodrugs has
`been developed further, and one class consists basically of quinones substituted with
`a side chain bearing a leaving group (Fig. 19). Following reduction to the hydro-
`quinone, quinone methide formation can result by elimination of HX. The quinone
`methides can then cause cytotoxicity by reacting with cellular nucleophiles (Fig. 19).
`This reduction to the hydroquinone, and hence activation, should take place more
`readily in hypoxic tumour cells.
`Further studies in this area have revealed that large numbers of naturally occur-
`
`ring quinones with biological activity may in fact be functioning as cytotoxic pro-
`drugs activated under conditions of low oxygen tension [49].
`
`
`
`
`
`hypoxic. Hypoxic cells tend to be more resistant to both radiation and cytotoxic
`therapy and may survive treatment which successfully kills the malignant oxy-
`genated cells [45]. These hypoxic cells may then be responsible for reforming the
`tumour after therapy. It has been suggested that a combination of cytotoxic agents
`with different targets (oxic cells, hypoxic cells, non—proliferating cells) may be much
`
`more successful than conventional chemotherapy [46]. That prodrugs may selective-
`ly attack hypoxic cells has been elegantly demonstrated by Frako and Chapman
`[47]. Nitroimidazoles are not cytotoxic in their own right but may be converted to
`their nitroradical anions and other reduced cytotoxic metabolites which bind
`covalently to cellular macromolecules. Using radioactively labelled nitroimidazoles
`and measuring covalent binding by autoradiography, the nitroimidazole prodrugs
`have been shown to be activated exclusively by cells which lie between the outer pro-
`liferating layer of solid tumours and the inner ischaemic zone.
`A number of types of prodrugs other than the nitroheterocyclics have been syn-
`thesised [48]. One of the first nitrogen mustards designed to be activated by reduc-
`
`tion was nitromin (Fig. 18). Nitromin is the N-oxide of nitrogen mustard (HZN) in
`which the lone pair of electrons takes place in bonding and can no longer activate
`
`
`
`
`
`Petitioner Mylan Pharmaceuticals Inc. - Exhibit 1012 - Page 158
`
`

`
`
`
`3 07
`
`For some alkylating agents, such as aromatic epoxides, electrophilic reactivity or
`antitumour activity can be influenced by the presence of neighbouring hydroxyl
`groups. It is known, for example, that the syn isomer of benzpyrene diol epoxide
`is some 163 times as reactive to 4—nitrothiophenolate than its corresponding anti-
`isomer [52]. The increased electrophilicity of the epoxide in the syn configuration
`is due to the anchimeric assistance of the cis hydroxyl group which probably, by
`hydrogen bonding, facilitates the opening of the epoxide ring. In the same way the
`drug triptolide is a good antileukaemic agent because it has a hydroxyl group cis to
`an epoxide. If the OH group is trans, all activity is lost [53]. Structures of the type
`shown in Figure 21 might be prodrugs if the ester group were a substrate for a
`tumour-specific esterase. The epoxide might bemstable and non—cytotoxic but
`hydrolysis of the cis ester would generate a hydroxyl group, which would convert
`the prodrug into a chemically reactive and cytotoxic species.
`
`5.3. ACTIVATION BY 7—GLUTAMYL TRANSFERASE
`
`High levels of 'y—glutamyl transferase occur in kidney tubules, and this has provided
`the basis
`for
`the
`synthesis of prodrugs
`activated by
`this
`enzyme.
`7-
`Glutamyldopamine, for example, causes a more potent and protracted elevation of
`serum glucose on infusion than dopamine [54], while 'y-glutamyl derivatives of
`
`306
`
`5.2. ACTIVATION BY ESTERASES
`
`Early histochemical studies using a variety of substrates have led to the claim that
`different tumour types (usually transplanted animal tumours) differ from ‘normal’
`tissues, either having a high esterase activity or sometimes lacking the ability to
`hydrolyse a specific substrate. No methodical survey has been reported of the dif-
`ferent types and levels of esterases in human tumours and the comparative values
`in normal tissue, so there has been no really rational design of prodrugs which might
`be activated by esterases. That selectivity may be achieved by this method is in-
`
`the benzoate and acetate of p-
`dicated from early work which showed that
`di—2-chloroethylaminophenol were more effective against a rat tumour than was the
`parent compound, although this might just as easily have been due to differences
`in tissue distribution [50]. However, enzyme activity does determine the antitumour
`selectivity of some benzoate esters. Thus, a 2—methylbenzoate ester had a better
`
`therapeutic index in tumour-bearing animals than the parent compound, whereas
`the 2,5-dimethylbenzoate ester had very little activity. This was related directly to
`the esterase activity [51], suggesting the prodrugs had little antitumour activity in
`their own right but required activation by tumour esterases to the active drug (Fig.
`20).
`'
`
`Compou nd
`
`'
`
`Ln
`
`50
`
`W90
`
`Enzyme
`activity
`
`H -M
`
`co.o O M
`
`on,
`
`CH3
`
`CH3
`
`M=N(CH2CH2)Cl2
`
`TI =Therapeutic Index
`
`Enzyme‘ activity = Phenol
`
`released iumol/25m x 10 '2)
`
`Fig. 20
`
`01
`
`\‘ OR
`
`L Esterase
`
`0;’
`
`‘~ OH
`
`1 Nu
`
`\ OH
`
`H0-
`
`Nu
`
`Fig. 21
`
`.
`
`Petitioner Mylan Pharmaceuticals Inc. — Exhibit 1012 — Page 159
`;'f“»\§(\
`”‘ “
`
`Petitioner Mylan Pharmaceuticals Inc. - Exhibit 1012 - Page 159
`
`

`
`
`
`309
`
`Plasminoqen
`
`Piasminoqen activator
`
`Plasmin
`
`D-VAL-EU-LYS-NH M
`
`M=NlCH2CH2Ci)2
`
`Fig. 23
`
`Normal chicken embryofibroblasts have a low level of plasminogen activator, but
`this increases dramatically if the cells are transformed by Rous sarcoma virus.
`Testing of the p-phenylenediamine peptidyl prodrug showed that it was at‘ least 5
`times more selective in its action against transformed cells when compared with p-
`phenylenediamine mustard. AT125 is an anticancer agent which also contains an
`essential primary amino group and formation of the peptidyl derivative of this, in
`a similar way, gave a prodrug that was specific in its cytotoxic action in vitro against
`
`fibroblasts producing plasminogen activator. However, in tumour—bearing animals
`there was no evidence of an increased therapeutic index, despite the fact that the
`prodrugs were tested against two transplanted tumours known to produce moderate
`to large amounts of plasminogen activator [58].
`A more promising prodrug appears to be the 3’—(D—Val—Leu—Lys) derivative of
`doxorubicin. This prodrug was
`similarly very selective against
`transformed
`fibroblasts in vitro and showed some evidence of activity on mice bearing the B16
`melanoma. The prodrug was a poor substrate for plasmin and did not allow an
`estimation of a therapeutic index [59]. A prodrug in which the peptidyl portion is
`attached to the doxorubicin group through an appropriate linker may greatly in-
`crease the prodrug’s susceptibility to plasmin.
`
`5.5.
`
`PRODRUGS ACTIVATED BY SEQUENTIAL PATHWAYS
`
`It is probably not likely that tumour cells will possess unique enzymes, and in the
`study of the plasmin—activated prodrugs, for example, there was evidence in the
`whole animal studies that the prodrug was also being activated in extratumoral sites.
`If, instead of one activation step being required, a prodrug is designed which can
`only be activated after two or more sequential steps, then it is more likely that there
`will be large differences between tumours and normal tissues. An early attempt to
`increase the number of steps required for prodrug activation was the synthesis of
`N,N—bis(2—chloroethyl)~N~glycyl—p—phenylenediamine mustard [60]
`(Fig. 24). Ac-
`tivation of the prodrug is a two—stage process. In the first step an acylase converts
`
`‘.
`
`Petitioner Mylan Pharmaceuticals Inc. — Exhibit 1012 — Page 160
`
`308
`
`sulphamethoxazole have also been synthesised as kidney—selective prodrugs [55]. A
`high level of this enzyme also occurs in some carcinogen—induced liver cancers in
`
`animals and in some human cancers [56, 57]. Although levels of the enzyme are high
`in the kidney, 7—glutamyl prodrugs may nevertheless be selective anticancer agents
`
`since kidney cells are inherently resistant