J. theor. Biol. (1998) 195, 111–128Article No. jt980792

0022–5193/98/021111+18 $30.00/0 7 1998 Academic Press

A Proposal for a New Direction to Treat Cancer

S R

Oncologic, Inc., 5920 San Pablo Avenue, Oakland, CA 94608, U.S.A.

(Received on 16 April 1998, Accepted in revised form on 13 July 1998)

A new approach is proposed that has the potential to be a successful therapy for mostdisseminated cancers because it can circumvent the problems posed by three characteristicswhich are universally expressed by cancer cells: heterogeneity, plasticity, and the lack of acancer specific or cancer associated characteristic which is not also shared by some normalcells.

Analysis shows that almost all current and research approaches for treating disseminatedcancers have the same fundamental strategy: they rely on an agent interacting individuallyand effectively with each cancer cell. We call all these approaches ‘‘lock and key’’ strategiesto emphasize the need for this individual agent to cell interaction. The three characteristicspreclude current approaches from successfully treating most disseminated cancers becausethey operate by a ‘‘lock and key’’ strategy which (a) only kills cancer cells expressing a singleparticular trait, (b) allows other cancer cells to adapt and survive the treatment, and (c) alsokills the normal cells which express the same particular trait.

The heterogeneity and plasticity of cancer cells can only be circumvented by an attackwhich is microregional (not cell by cell) and destructive (not killed by conventionalendogenous or exogenous cytotoxic agents). All cells in each microregion must be destroyed,including those which do not express an exploitable trait. The proposed approach can achievesuch microregional destruction by the delivery to, and long term immobilization of, a largenumber of radio-isotopes.

The proposed approach exploits the additive contribution of multiple mechanisms toenhance tumor specificity of the microregions. Given that all targeting and killing agents are‘‘imperfect’’, this is the only way specificity can be enhanced. The biological basis of thesespecificity enhancing mechanisms are well-known. However, they are ignored by currenttherapies because most of them can only be exploited in the context of the proposed approach.Some of the mechanisms reflect characteristics, such as heterogeneity, genetic instability, andtumor progression which are the result of the micro-evolutionary process of tumordevelopment. These are virtually always present in, and virtually specific to, cancer. Othersreflect the somewhat ‘‘imperfect’’ cancer associated characteristics of structures, includingcancer cells, extracellular structures, and non-malignant cells, within the tumor mass. Theadditive contribution of the multiple mechanisms gives the process the potential to destroyall the cancer cells with minimal non-tumor toxicity.

The cornerstone of the proposed approach is a novel class of soluble chemicals. They canbe administered intravenously to subjects, circulate throughout body fluids and areenzymatically converted into an insoluble material when the chemicals reach targeted sites.In this paper, these chemicals are called ‘‘soluble precipitable reagents’’ (SPR) to describetheir ability to be converted from a soluble to an insoluble material. The insoluble materialis called platform to indicate that it has the ability to bind various agents. The SPR chemicalsenable a three-step process to be constructed which can deliver and retain a large numberof radio-isotope atoms in tumor tissue.

. 112

In step 1, a binary reagent comprised of an SPR attached to an imperfect cancer targetingagent is administered. The binary reagent is endocytosed and transported into lysosomes wherethe targeting agentmoiety is digested and the detached SPR is converted by natural intracellularlysosomal enzymes into a platform. As will be discussed, a very large number of platformmolecules can be made to accumulate inside targeted cells.

In step 2, a supersensitive fraction of the cancer cells, including some which had accumulatedplatform in step 1, are killed by the administration of a very low dose of an anti-cancer agent.Very few, if any, normal cells will be killed by the very low dose. The death of the cells relocatesthe accumulated platform into the extracellular tissue fluid. After a lapse of time, any platformwhich is still retained in the extracellular tissue fluid can proceed to step 3.

In step 3, the platform is used by one of two different methods to generate supra-lethalradiation fields. This is achieved in the first method by the binding and long term retention ofa large number of isotope-carrying molecules to the extracellular platform. In the secondmethod, a bispecific reagent, having a non-mammalian enzymemoiety is bound to the platform.The bound enzyme converts a subsequently administered free (non-targeted) radioactive SPRinto an insoluble radioactive second precipitate which remains adjacent to the bound enzymefor a long time. Both methods, thus, result in the immobilization and long term retention ofa large number of isotope atoms in the tumor tissue, thereby generating intense radiation fieldswhich cause microregional destruction of thousands of neighboring cells.

7 1998 Academic Press

Introduction to the Cancer Problem

1. Problem posed by heterogeneity of cancer cells

Fifty years of intense research has shown thatthere is a wide heterogeneity in every character-istic or trait that has been measured in cancer cellpopulations (Rubin, 1990). One manifestation ofthe heterogeneity is that only a fraction of thecancer cells in any tumor population express atrait which makes them vulnerable to anyparticular agent. Cancer cells not expressing thistrait survive treatment by the agent and canbecome the dominant cell type in the tumorpopulation. No single agent can be expected totarget, treat or kill every cancer cell.

Principle which must be satisfied to circumventheterogeneity. Cancer treatment can only suc-ceed if it is capable of killing all cancer cells inthe tumor, including those that do not express anexploitable trait. Circumventing the problemsposed by heterogeneity requires the attack to bemicroregional and not cell by cell. Theoretically,it may be possible that this constraint need notapply if multiple agents were used to match theparticular and different traits present on eachsubset of the heterogeneous population of cancercells. The combinatorial diversity of the immunesystem (Weiss & Rajewsky, 1990) may have the

potential to generate multiple different agents tomatch the different traits. However, apart froma few cancers, for example some hematologicalcancers, this potential is unlikely to be realizedeven if the immune response is augmented byvaccines and cytokines. Some cancer cells simplymay not have an appropriate antigen to whichthe immune system can respond. Others may notexpress the antigen adequately through themajor histocompatibility complex. Malignantcells, particularly metastatic foci when comparedwith the primary tumor lesion, are frequentlyobserved to have absent or low expression ofthis histocompatibility complex (Cordon-Cardoet al., 1991; Elliott et al., 1989; Gopas et al.,1989).

2. Problem posed by plasticity of cancer cells

Cancer cells, like all organisms, can adapt andbecome resistant, via multiple genetic andepigenetic (Prehn, 1994) mechanisms, to manyuntoward environments which can include anexcess or deficiency of a natural agent in thesystem, or the addition of an artificial agent, suchas a cytotoxic drug.

Principle which must be satisfied to circumventplasticity. Circumventing the problem posed byplasticity requires the attack to be so intense thatit overwhelms all the cellular mechanisms,

113

including those responsible for adaptation. Inshort, the cells must be destroyed. As will bediscussed later in the paper, destruction of cellsis unlikely to be achievable by soluble drugs oragents, even if they are produced locally withinthe tumor. Destruction of cells, however, can bereadily achieved by immobilizing a large numberof radio-isotope atoms in each microregion andretaining them in their immobilized location foran extended time. A recent approach (which willbe described later) targets radiolabeled anti-bodies to extracellularly relocated DNA ofnecrotic tissues in an attempt to generate intenseradiation fields and achieve microregionaldestruction.

3. Problem posed by normal cells expressing thesame characteristics as cancer cells

Cancer cells in any one tumor do not expressan exploitable agent sensitive trait that is not alsoexpressed by some normal cells (or that is notexpressed in a form that prevents cross-reactionto the applied therapeutic agent). Unless aunique cancer specific trait is discovered, therewill always be some normal cells that express thesame trait as the cancer. This reflects theaccumulating evidence that most ‘‘cancer associ-ated traits’’ are not new functions, but rather, aredue to a dysregulation of existing normalfunctions. For example, it has been amply shownthat antigens expressed on melanomas reflect thedifferentiation state of their normal counterparts(Coulie et al., 1994; Houghton, 1994; Kawakamiet al., 1994). The host response may represent anapparent autoimmune recognition to differen-tiation antigens that are expressed as a directconsequence of the malignant state, or perhapsbecause the malignant state is ‘‘seen’’ as a formof tissue injury.

Principle which must be satisfied to circumventshared characteristics. Since there is no ‘‘perfect’’exploitable cancer trait or targeting mechanism,and since most scientists do not believe that onewill ever be discovered, it follows that the onlyway to improve specificity is by exploiting theadditive contribution of multiple ‘‘imperfect’’mechanisms. The degree of specificity that can beachieved depends on the number of mechanismsand the effectiveness of each one of them. Thepracticality of exploiting multiple mechanisms

requires that they do not put an additionalburden on the patient.

Disseminated cancer could be successfullytreated if highly radioactive, insoluble, andnon-digestible beads were accurately implantedin every cubic millimeter of the primary andmetastatic tumor tissue. An intense radiationfield would extend beyond each bead and destroythousands of cancer cells in the immediatemicroregion. In this paper we define intenseradiation fields and microregional destruction asfields which can destroy all cells in themicroregion, including those not expressing anyparticular agent sensitive trait. Accurate implan-tation would also prevent the therapy fromproducing significant non-tumor toxicity. How-ever, such an ideal implantation is not possibleby physical and surgical methods, and is the onlyreason why bead implantation is not a successfultreatment for disseminated cancers. On the otherhand, soluble radioactive agents are distributedthroughout the body fluids and can reach thetumor tissue virtually no matter where it islocated. However, unlike implanted beads,soluble radioactive agents cannot achieve a highenough concentration in any one microregion,and are not retained for long enough, to generatethe necessary intense radiation fields to cause celldestruction.

Logic demands that a therapeutic processhas the advantages of both soluble agentsand insoluble beads. Such a logic is satisfiedduring the successful treatment of disseminatedthyroid cancer by radio-iodide. Radio-iodide is asoluble agent which is distributed throughoutbody fluids, and is converted into a quasi-insoluble radioactive bead (colloid) by normaland malignant thyroid cells which enables itto be retained for a long enough time togenerate intense radiation fields. Since thetreatment of thyroid cancer faces the samecancer driven obstacles as any other cancer, andyet can often be treated successfully, it suggeststhat the same logic should be used to treat othercancers.

The proposed approach uses a novel chemicalto construct a three-step process which mimics,for non-thyroid cancers, the treatment of thyroid

Step 1Continued administrationof SPR binary reagent

Step 2Administer very low doseof cell killing lysing agent

Endocytosis

Enzyme digests targetingagent & converts SPRinto platform

Platform accumulationinside cells increaseswith time

Number of platformmolecules can be over1000x greater thanthe number of receptors

Cell membrane integrity lost

Targeting moietyspecific for binding

site on SPR

Each platform binds andretains a large numberof isotope carrying molecules

Generates supra-lethalradiation fields thatextend beyond eachplatform to causemicroregional destructionof thousands of cells

A large number of platformmolecules from each cellare relocated to the extracellular tissue fluid

SPRchemical

Targetingagent

Binding siteon SPR

Step 3 (first method)Administer isotopecarrying molecules

. 114

cancer with radio-iodide. Like radio-iodide,these novel chemicals are soluble when adminis-tered and precipitate by the catalytic action ofenzymes in targeted cells. In this paper, thesechemicals are called ‘‘soluble precipitable re-agents’’ (SPR) to describe their ability to beconverted from a soluble to an insolublematerial. The insoluble material is called aplatform to indicate that it has the ability to bindvarious agents.

The Proposed Three-step Approach

Current therapies for disseminated cancer usesoluble agents which have an immediatepharmacological effect on cells. In contrast, thestrategy of the proposed approach is a three-stepprocess which uses the SPR chemicals to createa biologically neutral, or relatively neutral,insoluble platform that is later used to createintense radiation fields by immobilizing and

F. 1. Three-step oncologic process.

115

retaining a large number of radio-isotope atomsin tumor tissue. The three-step process isdepicted in Fig. 1.

Step 1: a soluble binary reagent is administeredintravenously to cause the accumulation of aninsoluble platform inside targeted cells.

The soluble binary reagent is comprised of thenovel SPR chemical attached, either directly orvia a lysosomal enzyme sensitive link, to aprotein targeting agent for cancer cells.(Although the so-called targeting agents havesome preferential ability to target cancer cells,they do not target all cancer cells and they alsotarget some normal cells. The way in which theproposed approach circumvents this imperfec-tion is described later.) The binding of the binaryreagent to the specific receptors on the targetcells activates receptor-mediated endocytosiswhich transports the binary reagent into theacidic, enzyme rich, lysosomal compartmentsinside the cell (Geuze et al., 1986; Murphy, 1988;Pearse & Bretscher, 1981) where the targetingagent is digested and the SPR is converted bylysosome enzymes into an insoluble platform.

Step 1 of the proposed approach is a minormodification of published results which describethe delivery of a wide variety of different agentsto the lysosomes of targeted cells using binaryreagents and exploiting endocytosis, The deliv-ered agents include cytotoxic drugs, toxins, dyes,antidotes to toxic drugs, non-digestible carbo-hydrates and molecules carrying radio-isotopes(Wu et al., 1983, 1985; Firestone, 1994;Rushfeldt & Smedsrod, 1993; Pittman et al.,1983; Jansen et al., 1993; Pittman & Steinberg,1978; Pittman et al., 1979a, b). In all thesepublished examples, the detached chemical in thelysosome remains soluble. In the proposedapproach, the detached SPR is converted into aninsoluble platform.

There are two classes of SPR chemicals. Thefirst class are inherently soluble. When they aredetached from the targeting agent by lysosomeenzymes, a second set of lysosome enzymesconverts them into highly reactive intermediateswhich dimerize and create an entirely newmolecule which is highly insoluble—no solubliz-ing agent is required for this class of SPR. Thesecond class of SPR chemicals are inherently

insoluble, but are solublized by their attachmentto the protein targeting agent, and in some casesalso by their attachment to a solublizingpolymer. In the latter case, the solublizingpolymer is either digestible, or is made of smallsegments of non-digestible polymers with inter-vening digestible segments so that the smallnon-digestible segments can escape from the cell.A number of candidate SPR chemicals areavailable in each class, and apart from theirantigenicity, they have similar characteristics.This enables the proposed approach to sequen-tially use different SPR chemicals to avoidpotential problems which might arise should thesubject develop an immune response to any oneof them, even though they are only administeredfor a short time and even though, unlike mostother approaches, the treatment process of shortduration.

The candidate platforms made from the SPRchemicals are insoluble in both water and lipids,are relatively neutral to living cells, can act as anaffinity matrix by having binding sites, and canbe made to have an ordered structure, such as alinear polymer or micelle, so that the bindingsites are spaced at defined distances to reducesteric hindrance. The binding sites can be chosenfrom a large library of candidates, and can beartificial (not found in mammalian fluid or cells)to ensure that the complementary ligand canbind to them with high affinity and specificityand with minimal binding to natural structuresin the body.

For the following reasons, a very large numberof platform molecules can be made to accumu-late inside cells: (a) the accumulation occurs viareceptor-mediated endocytosis, which is a natu-ral cellular process, (b) the accumulationproceeds via a time dependent, cumulativeprocess, which can continue to operate as long asthe SPR binary reagent is available, (c) theplatform itself is non-digestible, and stable,which enables it to be retained (like the insolublematerial in tattoos) inside the cell for a long time,and (d) the platform is insoluble and relativelynon-toxic, which prevents it from osmotically orpharmacologically interfering with the ability ofthe cell to continue the accumulation process.The SPR can be prepared with a trace label insuch a way that the platform retains this label.

. 116

The long intracellular retention time of theplatform enables the amount and location of itsaccumulation to be followed by scan and/orbiopsy when there is a zero radiation back-ground in the body fluids.

Step 2: a very low dose of an anti-cancer agent isadministered which kills a small fraction of thecancer cells that are super-sensitive.

The death of these cells relocates theaccumulated platform to the extracellular tissuefluid where it is retained. The dose of the agentis so low that very few (if any) normal cells arekilled. As will be discussed later, this selectivekilling (together with platform accumulation) isone of the main mechanisms which determine thelocation of the final anti-cancer attack. It is notthe main factor in eliminating the majority of thecancer cells. After a period of time is allowed tolapse, platform which is retained in theextracellular tissue fluid is available to proceed tostep 3.

Step 3: the platform in the extracellular fluid isused by either of two different methods as alaunching pad from which the intense radiationfields are generated.

In the first method, radiation fields aregenerated by binding a large number of cellimpermeant isotope-carrying molecules to theartificial binding sites on the platform. In thesecond method, a bispecific reagent, having anon-mammalian enzyme moiety, is bound to theplatform. The bound enzyme converts a sub-sequently administered free (non-targeted) radio-active SPR into an insoluble radioactive secondprecipitate which remains in situ adjacent to thebound enzyme for an extended period of time.Both methods of step 3, thus, generate intenseradiation fields. The resultant microregionaldestruction circumvents the problems posed bythe heterogeneity and plasticity of the cancercells because all cells in the microregion aredestroyed irrespective of their biological status.

The binding of the isotope carrying moleculeand the bispecific reagent to relocated extracellu-lar platform is analogous to the in vivo bindingof radiolabeled antibody to extracellularlyrelocated DNA (Chen et al., 1990), myosin(Khaw et al., 1984 ), and cytokeratin (O’Brien &

Bolton, 1995) when the cells were killed.However in these published results, the antibodybinds to a natural, digestible structure of thebody and is removed by natural processes. Incontrast, the agents administered in step 3 bindto the sites on the artificial platform which isstable and non-digestible and for this reason thebinding can be more specific, have a higheraffinity, and can remain bound for a much longerperiod of time. This more stable binding isanalogous to the stable binding of antibodies toslow turn-over extracellular structures such asthe glomerular membrane of the kidney wherethe antibody remains bound for months (Yong& Rhodes, 1990).

The Proposed Approach Mimics the SuccessfulTreatment of Thyroid Cancer with Radio-iodide

Even though some thyroid cancer cells do nothave the trait that enables them to take up andstore the radio-iodide therapeutic agent, thetreatment of thyroid cancer with radio-iodide issuccessful in a high percentage of cases(Adelstein & Kassis, 1987; Fitzgerald et al.,1950). Since the treatment of thyroid canceressentially faces the same problems as othercancers and yet can often be successfully treated,the proposed approach was designed to mimicthe operating conditions that operate during thistreatment. The SPR technology enables thethree-step process to replicate, for non-thyroidcancers, the operating conditions that enable thetreatment of thyroid cancer to succeed.

1. The process is similar

In both the thyroid and the proposedapproach, a soluble agent is administered whichdistributes throughout the body fluids and canreach the disseminated cancer tissue where it isconverted via multiple steps and naturalintracellular enzymes into a relatively stablematerial (slow turn-over colloid in the thyroid,and insoluble platform in the proposed ap-proach). In both cases the accumulationproceeds in a time dependent cumulative mannerand can continue for as long as is necessary. Inthe thyroid model, the accumulated colloid isimmediately radioactive. In the proposed ap-proach the platform is not immediately radio-

117

active (except trace labeled) and only laterbecomes radioactive by either of the twomethods described for step 3 of the process.

2. The quantitative parameters that achievemicroregional destruction are similar

The ability to achieve intense radiation fieldsand microregional destruction is determined bythe number of radioactive isotope atoms whichare deposited in each microregion and theirretention time in this location. Calculations (inpart from preliminary data) show that the firstand second method of step 3 can achieve similarconditions as those which enable the treatmentof thyroid cancer to create intense radiationfields.

In the thyroid model, assume that there are 106

thyroid cancer cells per cubic millimeter, thatapproximately 50% of the cancer cells can takeup the radio-iodide, and that each of these cellscan take up and convert 104 radio-iodide atomsinto a colloid. There will be 5×109 isotopeatoms per cubic millimeter, thereby generating aradiation field of approximately 250 mCi pergram of tissue. The time taken for this uptakeand conversion is approximately 2–4 hr, oncetaken up, the isotopes are retained with abiological half-life of a few days. Since isotopecontinuously leaves the cancer, systemic toxicityis produced both during the uptake and releaseprocess.

In the proposed approach, the number ofisotope atoms which are ultimately depositeddepends on how much platform is accumulatedinside cells in step 1 and how many isotopeatoms are immobilized in step 3. As discussed, avery large number of platform molecules can bemade to accumulate inside cells in step 1. Theaccumulation of the platform reflects thememory of a cell’s biological activity over time.

Preliminary results have shown that (a) thecontinued administration of the SPR binaryreagent results in the accumulation of a largenumber of platform molecules inside targetedcells, (b) the process of forming and accumulat-ing the platform inside cells was relatively nottoxic, and (c) the platform is retained inside cellsfor a very long time. The hepatoma cell line,HepG2, known to express the specific asialogly-coprotein receptor, was grown in tissue culture

medium. The HepG2 cells were cultured induplicate cultures for 10 days in plastic ware(Falcon) at 37°C under 5% carbon dioxide and95% air in medium containing 5 mM concen-tration of the lactosilated polylysine-SPR. At theend of the culture period, the cells were washedthree times in balanced salt solution andharvested. The cells were incubated with 0.1normal sodium hydroxide for 30 min at roomtemperature, then dissolved in liquid scintillationfluid, and finally centrifuged in 2 ml centrifugetubes. The insoluble material was seen as a smallpellet made up of small particles approximately0.1 mm in diameter. Calculations from thisexperiment showed that each targeted cellaccumulated 1000 times as many molecules ofSPR-insoluble material as the number of specificasialoglycoprotein receptors.

The results above show that over 10 days,targeted cells can accumulate, in their lysosomes,1000 times as many molecules of platform asthere are specific receptors on the surface of thecell at any given time. Assuming that there are106 cancer cells per cubic millimeter, that only5% of cancer cells both accumulate platformduring step 1 and relocate platform in step 2, thateach cell has 104 receptors (a conservativeassumption), and that the receptor recycle rate is5 per hour. Using these assumptions, calcu-lations show that each of these cells canaccumulate at least 107 platform molecules in200 hr (104 receptors multiplied by receptorturn-over rate of 5 per hour multiplied by 200 hrof accumulating time) and therefore there wouldbe 5×1011 platform molecules per cubicmillimeter of tumor tissue. Assume that only10% of the platform molecules can bind theagent administered in the first or second methodof step 3. In the first method, the platformswould bind 5×1010 isotope atoms per cubicmillimeter which will remain bound for longerthan the retention time of radio-iodide duringthe treatment of thyroid cancer. Given both ofthese parameters, the first method of step 3 hasthe potential to generate a radiation field that isat least 10 times as intense as in the thyroidmodel.

The second method of step 3 (compared withthe first) has several advantages: (a) the secondmethod can generate radiation fields that are

. 118

very much more intense because the number ofisotope atoms which are immobilized is pro-portional to the size of the platform (like the firstmethod) multiplied by the turnover number ofthe enzyme (the turnover number of an enzymeis the number of soluble molecules that theenzyme can convert into an insoluble precipitateper unit time); (b) the isotope atoms remain intheir immobilized location for a longer period.Even if the radioactive precipitate is phagocy-tosed by macrophages, it will be retained insidethe macrophage cells within the tumor mass. Itis likely that the movement of the ‘‘hot’’macrophages within the tumor will distribute theradioactivity more evenly throughout the tumorwhich would enhance the probability of destroy-ing more cancer cells in the tumor (macrophagesin tissues are not thought to return to the generalcirculation, though they can move to the regionallymph glands). In contrast, in the first method,if the platform with its bound isotope carryingmolecule is phagocytosed by macrophages, theattachment of the isotope to the platform will bedisrupted by the combined effect of low pH andthe action of lysosome enzymes. This will causethe isotope to escape from the cell therebydecreasing the intensity of the radiation fields inthe tumor and increasing the systemic radiationtoxicity; (c) furthermore, less systemic toxicity islikely to be produced by the second methodbecause it is much easier to make a smallradioactive SPR which can be converted into aprecipitate, compared with making a smallisotope carrying molecule which will bind to theplatform. The candidate radioactive SPR hasideal characteristics of pool distribution andexcretion kinetics because it is small and can onlybe converted into an insoluble precipitate by thecatalytic action of the non-mammalian enzymemoiety of the bound bispecific reagent. Ananalogue of this candidate SPR is known tocirculate in the blood without causing significantsystemic toxicity.

The radiation fields generated by the first, andparticularly by the second method, of step 3 canbe hundreds or even thousands of times moreintense than those produced when treatingthyroid cancer. The fields can be intense enoughto cause microregional destruction (radio-necrosis) of thousands of cancer cells in each

microregion irrespective of their biological statusand even if they have not proceeded throughsteps 1 and 2 of the process. In this way theproposed approach mimics the killing of thosethyroid cancer cells that do not take upradio-iodide by the radiation fields generated bythe thyroid cancer cells that do take up theisotope. Microregional destruction is the onlyway to defeat the problems posed by theheterogeneity and plasticity of the cancer cellpopulation, and it circumvents the necessity offinding a uniquely exploitable trait common toall cancer cells.

3. Achieving insignificant systemic radiationtoxicity

Even though radio-iodide atoms circulate inthe blood during uptake and are slowly releasedfrom the colloid during the treatment of thyroidcancer, only insignificant systemic radiationtoxicity is usually produced because radio-iodidehas ideal pool distribution and excretioncharacteristics. Although the isotope carryingmolecule used in step 3 is unlikely to have thesame ideal distribution characteristics as radio-iodide, this deficiency can be compensated for bythe other more advantageous parameters. It isunlikely that the proposed approach willproduce significant systemic toxicity for thefollowing reasons: (a) the large number ofbinding sites on the platform increases the rateof transfer of the isotope carrying molecule fromthe blood and its immobilization on, or adjacentto, the platform (binding to the platform in thefirst method and conversion to a precipitate inthe second method); (b) in both methods, thelong retention time of the immobilized isotopesincreases the ratio of tumor vs. systemicradiation dose—particularly if an isotope with arelatively long physical half-life is used; (c) inboth methods, the isotope carrying molecule canbe chosen from a wide variety of candidates tohave the best pool distribution and circulationkinetics.

4. Overlapping microregions of radiation

Overlapping microregions of destruction arerequired to eliminate all the cancer cells in thetumor. In the thyroid model, the microregionsoften overlap because many thyroid cancers are

119

sufficiently well differentiated that cancer cellswith positive iodide trapping traits are closeenough together so that the fields they generateoverlap. Achieving this necessary overlappingfor most solid tumors treated by the proposedapproach may require a second administrationof the isotope-carrying molecule which willproduce a second round of radiation fields. Forthe following reasons, the number of secondround radiation fields is likely to be much greaterthan the first round: (a) all cancer cells whichhave accumulated platform and which are in theouter peripheral rim of the first round ofradiation fields will be destroyed, and as aconsequence will relocate their platform to theextracellular tissue fluid of the rim; (b) there willbe no convective flow or viable macrophages inthe rim because blood capillaries, lymphatics,and macrophages in the microregion and rimwill, like the cancer cells, also be destroyed. Asa consequence, the platforms relocated in thetissue fluid of the rim will remain in situ for anextended time because they cannot easily‘‘escape’’ from the extracellular tissue fluid; (c)although insoluble particles and platform canonly move by convective flow, small solubleisotope carrying molecules can move by diffusioninto the convective free rim.

The ability of the proposed approach togenerate a second round of radiation fieldsgreatly increases the chance that the micro-regions will overlap and that the therapy willdestroy all the cancer cells in the tumor. One canview the few supersensitive cancer cells which arebiologically killed in step 2 of the process, andeven the cancer cells that are destroyed in thefirst round of radiation fields, as simply being‘‘markers’’ for the main destruction achievedduring the second round. Since thousands ofcancer cells are destroyed around each platformin the first and second round fields, and since thenumber of second round fields will be greaterthan the first ones, the therapy is notcompromised even if the number of cells thatboth accumulate platform in step 1 and aresupersensitive and killed in step 2, is small. Thepractical ability to carry out second roundradiation fields depends on producing a low levelof systemic radiation toxicity in the first round,and this in turn, largely depends on the

characteristics of the isotope carrying moleculeand how rapidly these molecules are immobi-lized. Second round fields are analogous to thesecond administration of radio-iodide which isoccasionally necessary to treat thyroid cancer.

5. The thyroid model and the proposed approachachieve tumor specificity in different ways

In the thyroid model, the radiation fields arelocated specifically and exclusively in normal andmalignant thyroid tissue because radio-iodide isa ‘‘near perfect’’ and exclusive targeting agent forthis tissue. For this reason, the proposedapproach cannot mimic the method by which thethyroid model achieves the location specificity ofthe radiation fields.

Tumor Specificity of the Proposed Approach

Since there is no equivalent ‘‘near perfect’’targeting mechanism for non-thyroid cancer(most scientists do not believe that such amechanism will ever be discovered) the proposedapproach uses the additive contribution ofmultiple mechanisms to improve the tumorspecificity of the radiation fields. These mechan-isms, which do not put a toxic burden on thepatient, operate in each step of the process tomaximize the differences in the number and sizeof the platforms in tumor tissue vs. normal tissueprior to step 3.

The difference in the amount of platformavailable in step 3 is determined by where andhow much platform is accumulated in step1,where and how much is relocated to theextracellular tissue fluid in step 2, and where andhow much is retained in the extracellular tissuefluid prior to step 3. Finally, for any givenplatform distribution, the specificity of theradiation fields is determined by the specificity ofthe isotope immobilizing process.

1

Preferential accumulation of platform can beachieved in cancer cells that over-express highaffinity binding endocytosing receptors. Themicro-evolutionary process and tumor pro-gression (Cheng & Loeb, 1993; Foulds, 1969,1975; Hill, 1990) make it likely that cancer cells

. 120

having such receptors for growth promoting orsurvival ligands will become a dominant celltype.

The preferential accumulation of platform canbe further enhanced if the binary reagent isadministered at a continuous low dose infusionover time. This simulates the low concentrationof natural circulating ligands in body fluidswhich is a major factor contributing to theirspecificity. The low dose of the binary reagent isanalogous to the phenomenon of low doseadministration of antigen resulting in theimmune system generating high binding anti-bodies because the antigen preferentially bindsto, and causes the proliferation of, immune cellsexpressing high binding receptors for theantigen.

The low concentration is feasible in thecontext of the proposed approach because theplatform can be made to accumulate slowly andis only used later when sufficient platform hasaccumulated and has been retained. In contrast,the agents used in lock and key strategies mustbe given at a high dose to achieve a relativelyhigh concentration so that they will exert theirimmediate pharmacological action.

2

A very low dose of an anti-cancer agent isadministered in step 2. This low dose will kill thesmall fraction of cancer cells which aresuper-sensitive and causes the SPR-platform tobe relocated to the extracellular tissue fluid. Thedose of the agent is so low that very few, if any,normal cells will be killed. The goal of theselective killing (together with platform accumu-lation) is to be one of the markers for thelocation of the final anti-cancer attack. It is notthe main factor in directly killing the majority ofthe cancer cells.

The presence of these super-sensitive cells in atumor, and not in normal tissues, reflects thegenetic instability (Loeb, 1998) and resultantheterogeneity (Rubin, 1990) of the cancer cells(features which are considered to be thehallmarks of the disease). The presence ofsuper-sensitive cancer cells may be one of themost common features of tumor populations,but they are totally ignored by current research,

presumably because these ‘‘easy to kill’’ cells areof no value in the context of lock and keyapproaches in which the problem of resistant andmulti-drug resistant cells is a critical problem.However, when used in the proposed approachas a tumor specific marker determining thelocation of the anti-cancer attack, the presenceand ‘‘easy killing’’ of these super-sensitive cancercells may be the most exploitable feature ofcancer. In fact, the administration of low dosesof current therapeutic agents can be consideredas a non-toxic enhancement of the ongoingnatural killing of super-sensitive cancer cells bythe defense systems of the body (Wyllie, 1992).

In the proposed approach, the tumor specifi-city depends primarily on the radiation fieldsbeing generated only if the same cancer cell bothaccumulates platform in step 1 and relocates theintracellular platform to the extracellular tissuefluid in step 2. This double requirement shouldenhance specificity because it requires the samecell to have two different and independent traits.The first trait is the expression of endocytosingreceptors that enable the cell to accumulateplatform, and the second trait is the cell’ssuper-sensitivity to being killed by low doses ofanti-cancer agents which relocate the platform tothe extracellular tissue fluid. As described, thetherapeutic process is not compromised if thereis a very low number of cells with this double setof characteristics because it is possible togenerate a second round of radiation fields.

The agents used in step 2 of the proposedapproach cause lysis, which destroys the integrityof the cell membrane and results in theintracellular contents, including the platform, tobe relocated to the extracellular tissue fluid(which is the only location where the platformcan generate radiation fields). Therefore, lysedcells can generate radiation fields. In contrast,the death of normal cells which occurs byapoptosis as part of their natural cellularturn-over does not cause relocation of theplatform to the extracellular tissue fluid.Apoptosis in vivo results in the formation ofimpermeable membrane-bound vesicles whichare rapidly engulfed by adjacent parenchymal orprofessional phagocytes (Stewart, 1994; Wyllie,1992). This prevents the release of theirintracellular constituents, including accumulated

121

platform, into the extracellular space andprevents radiation fields from later beinggenerated. Radiation fields are not generatedeven if normal cells have accumulated platformand have died as a result of their natural cellturnover. Thus, apoptotic death of normal cellswill tend not interfere with the tumor specificityof the radiation fields.

3

Normal epithelial cells exfoliate into theexternal environment when they die or are killed(Croitoru & Riddell, 1993). This exfoliation hasbeen demonstrated for epithelial cells of thegastrointestinal tract (Ishikawa et al., 1993),kidney (Goligorsky et al., 1993; Walker, 1994),bladder (Rebel et al., 1994), lung (Bogdanffy etal., 1994; Finotto et al., 1993), and others.Exfoliation causes dead normal epithelial cellsand their intracellular contents to be dischargedinto the external environment where theyobviously cannot generate radiation fields.Therefore, even if any normal epithelial cellsaccumulate platform in step 1 (because they havethe same endocytosing receptor as the cancer)and are killed in step 2 (because they weresuper-sensitive), they will not produce radiationfields.

In contrast, epithelial cancer cells (with a fewexceptions, such as superficial bladder cancers),are not located on the boundary between theinternal and external environment and theycannot exfoliate into the external environment.When these epithelial cancer cells are killed, theyrelocate their accumulated platform to theextracellular tissue fluid (this being the onlylocation that enables radiation fields to begenerated).

Similarly, killed endothelial cells are dis-charged into the blood stream and not into theextracellular tissue fluid (Baillie et al., 1995;Yuzawa et al., 1994). Once in the blood stream,these cells and any platform they had previouslyaccumulated, will be quickly engulfed by thephagocytic systems of the lung, liver, andspleen. In this intracellular location, the platformwill not be able to generate radiation fields.

It should be noted that endothelial celldamage and death are common and seriousside-effects of immunotoxins (Soler-Rodriguez etal., 1993) and of interleukin-2 therapy (Fujita etal., 1994).

The following example demonstrates thecontribution of epithelial cell exfoliation andendothelial cell discharge into the blood streamin determining the tumor specificity of theproposed approach. An immunotoxin was madeby attaching a pseuodomonas exotoxin A (PE)with its cell receptor binding portion deleted, toan antibody, MAbB3 to make a binary reagent.The MAbB3 binds to 90% of metastatic coloncancer but does not react with normal colon. Italso reacts with other metastatic cancersincluding many breast, ovarian, esophageal,stomach, bladder and prostate cancers. Theantibody is not entirely cancer specific and reactswith a limited number of normal epithelialtissues including, glands of the stomach, thesuperficial epithelium of bladder, trachea, andesophagus. The immunotoxin could causetoxicity in these normal tissues. However, in theproposed approach, the same MAbB3 antibodycould be used to make the SPR binary reagentto accumulate platform in step 1. As describedabove, even the platform accumulation occurredin these normal cells, and even if some of thesenormal cells were killed in step 2, the specificityof the therapy would not be compromised.Similarly the antibody portion of the immuno-toxin may also react with B3 antigens onendothelial cells and cause vascular leaksyndrome. However, as described above, theproposed approach would not be compromisedbecause killed endothelial cells are dischargedinto the bloodstream and the released platformwould be quickly phagocytosed by the macro-phages in the liver, lung and spleen. Once insidethese macrophages, the platform would not beable to generate a radiation field in step 3. Areview of immunotoxins has recently beenpublished which describes and references manytargeting antibodies that have been used asbinary reagents to deliver toxins to cancers(Pastan,1997). It should be noted that many ofthese antibodies could also be used to deliver theSPR without causing the difficulties experiencedwhen they are used to deliver toxins.

. 122

In addition, platform which has been relocatedto the extracellular fluid of cancer tissue will beretained in this location for a longer periodcompared with the extracellular fluid of normaltissue. This difference in retention time creates atemporal window that can be exploited toenhance the tumor-specificity of the radiationfields by appropriately timing the initiation ofstep 3.

The lack of lymphatics in cancer tissueexplains why trypan blue (which binds toalbumin) injected into tumor tissue is retained insitu for up to 5 days, compared with only hourswhen injected into normal tissues (Matsumura &Maeda, 1986). Given the longer retention of thesoluble macromolecules in this example, it can bepredicted that, compared with normal tissue, theretention time of the insoluble platform in thecancer tissue is likely to be much longer.Convective flow in normal tissues transportsinjected particles into the lymphatic channelsand then to the regional lymph glands where theyare effectively engulfed by the resident macro-phages lining the sinusoids (Ikomi et al., 1995).By analogy, platform that has been relocated tothe extracellular tissue fluid will also betransported to lymph glands and be engulfedinside the macrophage cells where the nowintracellular platforms cannot generate radiationfields. In contrast to normal tissues, the lack oflymphatics in cancer tissue (Fallowfield & Cook,1990; Jain, 1987; Seymour, 1992) reduces thepossibility of this transportation and engulfmentprocess, and as a consequence, the platforms willbe retained in cancer tissue for a much longertime than in normal tissue.

All the specificity enhancing mechanismsdescribed above operate without adding anyextra burden to the patient and withoutadministering any additional agents. The pro-posed approach uniquely benefits from thebody’s natural functions, and particularly thosefunctions which have been altered by thepresence of the tumor. The combined effect ofthese multiple mechanisms results in the presenceof a very large amount of platform in theextracellular fluid of tumor tissue, and very little,if any, platform in the extracellular fluid ofnormal tissues at the time the isotope carryingmolecule is administered in step 3.

3

Specificity is not degraded by the first methodof step 3 because the binding sites on theplatform are foreign to mammalian structuresand the ideal binding sites can be chosen from awide variety of candidates. For these reasons, itis possible to make the isotope carrying moleculeso that it will bind with high specificity andaffinity to the sites on the platform, withoutbinding significantly to cells or structures in thebody. As a consequence of this exclusive binding,the radiation fields will be generated only aroundthe platforms. Since the platforms will be locatedpreferentially, and perhaps virtually exclusively,in the tumor, the radiation fields themselves willbe located virtually exclusively in the tumor.Specificity is not degraded by the second methodof step 3 because the radioactive SPR is onlyconverted into a radioactive precipitate by thenon-mammalian enzyme moiety of the bispecificagent, which itself is bound only to the platform.

-

In the proposed approach, the tumor specifi-city achieved by the natural mechanisms can begreatly enhanced by the administration of avariety of non-toxic agents. For example,targeting molecules specific for homogeneousreceptors (for example, cell lineage receptors N)on a set of normal cells can be used to deliver avariety of non-toxic agents that inhibit one ormore of the essential steps of the radiationgenerating process from operating. Because thenormal cells of any one tissue are relativelyhomogeneous and express N, they can betargeted and the radiation generating processcan be inhibited in most of these N positive cells.Specifically, the administration (prior to step 2)of an enzyme targeted to N can cleave thebinding epitopes on the platform inside thesecells. This would prevent the agents in step 3from binding to the platform and thus wouldprevent radiation fields from being generated.

123

Alternatively, these targeting molecules could beused to deliver, to N positive cells, an antidoteto the low dose of the anti-cancer agentadministered in step 2. This would protect Npositive cells from relocating their platform tothe extracellular tissue fluid in step 2, which inturn, would prevent radiation fields from beinggenerated around them in step 3. Used in thisway, targeting molecules could prevent theradiation generating process from operating inmost normal cells.

This could markedly reduce non-tumor tox-icity. Even if a few of the heterogeneous cancercells have the receptor N and the radiationgenerating process was inhibited in these few, thetherapeutic process would not be compromisedin its ability to destroy all cancer cells becauseeach radiation field destroys thousands of cancercells, and because the frequency of second roundradiation fields is much higher than first roundfields. Thus, the therapeutic improvementachieved by preventing radiation fields in normaltissues can be gained without compromising theability of the process to destroy all the cancercells.

The strategy of preventing radiation fields innormal tissues exploits the marked homogeneityof normal cells compared with the markedheterogeneity of cancer cells. Lock and keystrategies cannot use the strategy of protectingnormal cells because some cancer cells wouldalso be protected, and, without microregionaldestruction, such protection would be counter-productive. For example, folinic acid is used asan antidote to systemic methotrexate. This doesprotect normal cells and reduces systemictoxicity, but it is highly likely that the folinic acidwould also protect some cancer cells.

Specificity of the proposed approach can befurther enhanced because it can exploit virtuallyuniversal and specific characteristics of cancercells such as genetic instability (Loeb, 1998) andtumor progression (Foulds, 1969, 1975). Theproposed approach can exploit these character-istics, as described below, because the platformis stable and remains inside lysosomes for anextended period of time. At the beginning of thetherapeutic process, some cancer cells and somenormal cells express the endocytosing receptor X(which is responsible for the accumulation of the

intracellular platform) and the platform willcontain a binding site Y (to which the isotopecarrying molecule or the bispecific reagent in step3 can later bind, and which can only be cleavedfrom the platform by a non-mammalianenzyme). After the intracellular accumulation ofplatform is completed, and after a period of timehas been allowed to lapse to allow the geneticinstability (Loeb, 1998) and the ‘‘tumor pro-gression’’ of the cancer cells to be expressed,(Cheng & Loeb, 1993; Foulds, 1969, 1975; Hill1990; Loeb, 1998), it is likely that some cancercells will have arisen which have lost X.However, since normal cells are genetically andepigenetically stable they will continue to expressX. Prior to step 2, targeting agents which are Xspecific can be used (as described above) todeliver agents (enzymes, antidotes, etc.) to Xpositive normal cells and X positive cancer cellsbut not to the few X negative cancer cells. Eitherof these delivered agents will prevent radiationfields from being generated around X positivecells while still permitting radiation fields to becreated around X negative cells that wereoriginally X positive. Exploiting the uniquechange in the few cancer cells from being Xpositive to X negative will greatly increase tumorspecificity. Although the number of cells losingthe receptor X in a relatively short time is likelyto be very small, their location should be highlyspecific to the cancer.

The potential problem posed by the lowfrequency of radiation fields in the cancer can becircumvented by the generation of a largenumber of second round of radiation fields. Forexample, X can be used to deliver an agent whichprotects X positive normal cells from being killedin step 2. In this scenario, the binding sites on theplatform in cancer cells will be intact, the firstround radiation fields will destroy thousands ofcells which will relocate their platform to theextracellular tissue fluid. These relocated plat-forms will be able to generate second roundradiation fields. In this way, virtually all cancercells which had accumulated platform in step 1will be able to generate second round fields.Thus, tumor specificity and overlapping radi-ation fields are both achievable.

Lock and key strategies attempt to achievespecificity by finding an agent that acts effectively

. 124

on a large number of cancer cells and only on asmall number of normal cells. The proposedapproach is not limited by these constraints. Forexample, if five different agents are available totarget cancer cells or to kill them, the proposedapproach can select and use the agent even if itis specific to only a few cancer cells and even ifit is specific to a large number of particularnormal cells that cannot generate radiationfields. These latter normal cells include epithelialor endothelial cells that inherently cannotgenerate radiation fields, and/or those normalcells which have a common homogeneousreceptor, such as the common receptor on B cells(Scheinberg, 1995) or hepatocytes (Geuze et al.,1986), which can be targeted to inhibit theradiation fields from developing around them.Nonetheless, as described earlier, even if a fewcancer cells are also protected the finaltherapeutic outcome is not compromised.

The aggressiveness and tumor specificity ofany therapeutic process or approach is greatlyinfluenced by the various intra-tumor structures.The characteristics of intra-tumor extracellularstructures and non-tumor cells are influenced bythe characteristics of the cancer cells. Since thelatter are heterogeneous, it is likely that thecharacteristics of the former will also beheterogeneous. As a consequence, any process,agent, or mechanism that is effective in onelocation of the tumor may be ineffective inanother location of the same tumor. In order tokill cancer cells in sub-optimum locations, forexample areas of poor blood supply or having aparticular cytokine microenvironment, the thera-peutic process must be able to generate an attackthat is many times in excess of what is necessaryto kill cancer cells in therapeutically optimumlocations. The proposed approach can achieveradiation fields that are sufficiently intense tosatisfy this excess requirement as shown by thecalculations described earlier and similarly,‘‘excess’’ tumor specificity can be readilyachievable.

It is important to note the proposed approachexploits multiple mechanisms to enhance tumorspecificity in each step, some of the mechanismsoperate naturally and others require non-toxicintervention. Since the final outcome of thetherapy is the result of the individual effective-

ness and the additive contribution of themechanisms, the proposed approach has thepotential to be highly specific and to be capableof future improvements. This emphasizes one ofthe critically important features of the proposedapproach: multiple steps provide multiple oppor-tunities to improve the final result withoutputting a burden on the patient. In this way, theproposed approach is analogous to the multiplemechanisms the body uses to achieve a functionin a specific and controlled manner. In contrast,the therapeutic outcome of lock and keystrategies that operate by a single step cannot beimproved significantly.

Discussion

Three strategies have been developed whichattempt to circumvent the obstacles facing lockand key strategies. They operate by convertingtargeted cancer cells into ‘‘production sites’’ forsoluble cytotoxic agents which diffuse into theimmediate microregion and attempt to killneighboring non-targeted cancer cells. Thesestrategies include: antibody dependent enzymepro-drug therapy or ADEPT (Rodrigues et al.,1995; Smith et al., 1997), vector dependentenzyme pro-drug therapy or VDEPT (Trinhet al., 1995), and therapy by gene insertiondesigned to over-produce natural peptides(Weichselbaum et al., 1994). These threestrategies represent a recognition of the need forthe anti-cancer attack to kill cancer cells thatcannot be targeted. However, these approachesare actually all special cases of lock and keystrategies and for the following particularreasons they are likely to fail: (a) since thesoluble agents diffuse into the blood and cause alevel of systemic toxicity which is proportional tothe toxicity produced locally, the local concen-tration cannot be made to exceed a certain level.This level must be lower if the tumor is largebecause the larger the tumor, the greater thenumber of drug producing locations; (b) it islikely that some cancer cells will be resistant (andeven adapt to become super-resistant) to themarginally higher concentration of the locallyproduced cytotoxic agents; (c) the cells on whichthe agent is generated (ADEPT), or the cellswhich produce the cytotoxic agent (VDEPT),

125

will be exposed to the highest concentration ofthe agent and are likely to be among the first cellsto be killed which prevents further production ofcytotoxic agent; (d) the location of the attackcannot be made sufficiently specific because thefirst step of each of these strategies uses a single‘‘imperfect’’ targeting agent to convert them into‘‘production sites.’’

Like the proposed approach, a recent strategy,called tumor necrosis therapy (TNT), has beendeveloped which recognizes that it is impossiblefor any therapy to succeed if it depends on killingcancer cells which have a particular character-istic or trait. TNT also recognizes that destruc-tion of cells can best be achieved by immobilizinga large number of radio-isotope atoms in eachmicroregion of the tumor and retaining them intheir immobilized location for an extended time.The TNT binds a radiolabeled antibody toextracellularly relocated DNA of necrotic ordying tissue (Chen et al., 1989, 1990). Data fromclinical trials have produced encouraging resultsand have shown that TNT can achieve aradio-isotope uptake by the tumor 5–8 timeslarger than radiolabeled antibodies targeted tocell receptors as in conventional approaches.

Instead of using naturally present DNA as inTNT, the proposed approach uses an artificialplatform which has previously accumulatedinside targeted cells. The following analysissuggests that this seemingly minor differenceenables the anti-cancer attack in the proposedapproach to be much more aggressive, moretumor specific, and to have the potential to keepimproving:

(a) compared to TNT, the proposed approachhas the potential to deposit a larger number ofradio-isotope atoms in the tumor and to retainthem for longer. The number of radio-isotopeatoms which can bind to the DNA isproportional to the fixed amount of this materialin the cell. The retention time of the isotopes islimited because the DNA is a digestible material.In the proposed approach, the number ofisotopes which bind is proportional to thenumber of platform molecules, and as has beendescribed, this number can be made very largebecause it is proportional to a time dependentcumulative process which can continue until the

required number has accumulated. The isotopewill be retained for a longer time because it isbound to a stable, non-digestible platform. Boththese parameters give the proposed approach thepotential to generate more intense radiationfields than TNT;

(b) TNT antibody may cross-react with othernatural structures in the body because it binds tothe natural epitopes on the DNA. In contrast,the agents administered in step 3 of the proposedapproach bind to the artificial site on theplatform. Since these sites can be chosen from alibrary of candidates, the agent can bind to themwith a high affinity and specificity, and not bindto natural structures in the body;

(c) TNT uses a single ‘‘cancer associatedcharacteristic’’—namely the presence of dead ordying cells—to locate the radiation fields.Although dead cells are found preferentially intumor tissue, reduction in tumor specificity mayresult from the presence of dead or dying normalcells (which occur as part of their naturalturnover) throughout the body. The location ofthe radiation fields in the proposed approachrequires a cell to have two independentcharacteristics (the cell must have the appropri-ate receptor to accumulate platform and it mustbe supersensitive to being killed in step 2). Thisdouble requirement by itself enhances specificity.However, as described, the tumor specificity inthe proposed approach can be further improvedby the delivery of targeted agents (enzymes andantidotes) to homogeneous receptors on normalcells which will cleave the binding sites on theplatform contained in these normal cells orprotect them.

Although there have been significant advancesin molecular biology, in biotechnology, and inthe understanding of important cellular pro-cesses, they have not been translated into asuccessful (or even a greatly improved) therapyfor most solid cancers. In a 1997 article entitled‘‘Cancer Undefeated,’’ Bailar gave a statisticalanalysis of the results of cancer therapy overthe last decade. He asserted that there hasbeen virtually no improvement in the mortalityrate of most cancers (Bailar & Gornik, 1997)although there has been some advance intreating hematological cancers (Scheinberg,

. 126

1995), neuro-blastomas (van Noesel et al., 1997),and testicular cancers (van Basten et al.,1997).

Analysis shows that almost all current andresearch approaches for treating disseminatedcancers have the same fundamental strategy:they rely on an agent interacting individually andeffectively with each cancer cell via its appropri-ate agent sensitive trait and it is for this reasonthat they have failed to substantially improve.We called all these approaches ‘‘lock and key’’strategies to emphasize the need for thisindividual agent–cell interaction. They aredefeated by the problems posed by the threeuniversal characteristics of cancer cells describedearlier. These approaches include chemotherapy,immunotoxins, immunotherapy, signal transduc-tion, apoptosis, cell cycle control, and variousforms of gene therapy. The claim has been madethat the new approaches, based on modernmolecular biology, will be able to kill morecancer cells with less non-tumor toxicity. Thisclaim is unlikely to be true—by virtue of theiressential strategy, they will only kill a fraction ofthe cancer cells, therefore, there will be atendency (as in the past) for the therapist to keepincreasing the dose in an attempt to kill all thecancer cells. This higher dose causes the systemictoxicity common to all these approaches.

Despite the inherent limitations of lock andkey based strategies, ‘‘euphoric’’ announcementsare repeatedly made in the media and bycompanies for the latest of these strategies. Onlymuch later, the same strategy is shown to fail.For example, ‘‘magic bullets’’ were onceconsidered a major break-through, but are nowdescribed in articles as the ‘‘bullet that flopped.’’Although significant improvements were made inthe design and manufacture of these bullets, theimprovements did not address the fundamentalreasons why they originally failed so it is notsurprising that they continue to fail. A similarhistory is being repeated for the latest lock andkey strategies. For example, attempts to treatcancer by correcting P53 gene error, are likely tofail because the P53 gene error is not found inevery cancer cell, genes cannot be inserted into,or corrected in, every cancer cell having theerror, and other gene errors are also presentwhich contribute to the malignant state.

In 1998, Klein wrote an article from which twocomments are quoted. ‘‘Biochemists, immuno-logists, cytogeneticists and virologists have keptsearching for the decisive event that had to bediscovered before the riddle of cancer could besolved—a common metabolic disturbance, themirage of a common cancer antigen, theuniversal virus, a general chromosomal imbal-ance, are examples of the all-encompassingtheories that were advocated with particularfervor’’. This was followed by a later comment:‘‘it has been said that the investment into cancerresearch has been a waste, that science did notlive up to its expectations, and that cancersupports more people than it kills’’ (Klein, 1998).

A consideration of the biological basis of thethree problems facing cancer therapy, and of theneed for ‘‘excess’’ specificity enhancing andkilling potential reveals that the failure of lockand key approaches reflects their essentialstrategy (the mirage of a common metabolicdisturbance), and is not due to the exactcomposition or action of any therapeutic agent.The implication of this conclusion is inescapable:if these strategies continue to dominate the field,future efforts in cancer therapy will continue tofail, and Bailar will surely be able to writeanother of his ‘‘Cancer Undefeated’’ articles.

It is evident that a new direction isneeded—one that abandons the simple lock andkey strategies and embraces the complexities thatreflect the microevolutionary process by whichthe tumor develops and progresses. The pro-posed approach represents such a new directionand it has the potential to circumvent the manycancer driven obstacles facing the problem ofconstructing a successful therapy for cancer.

REFERENCES

A, S. J. & K, A. L. (1987). Radiobiologicimplications of the microscopic distribution of energyfrom radionuclides. Int. J. Rad. Appl. Instrum. (B) 14,165–169.

B, C. T., W, M. C. & B, N. J. (1995).Tumor vasculature—a potential therapeutic target. Br. J.Cancer 72, 257–267.

, J. P., S K, H., S, D. T.,P, E., D, M. F. & H, H. J. (1997).Current concepts about testicular cancer. Eur. J. Surg.Oncol. 23, 354–360.

127

B, M. S., D- M, H. C., B,R.B., F, V. J.,C, T.C., T, T.R.,V,M. B. & R, R. W. (1994). Chronic toxicity andoncogenicity inhalation study with vinyl acetate in the ratand mouse. Fundam. Appl. Toxicol. 23, 215–229.

C, F. M., T, C. R. & E, A. L. (1989). Tumornecrosis treatment of ME-180 human cervical carcinomamodel with 131I-labeled TNT-1 monoclonal antibody.Cancer Res. 49, 4578–4585.

C, F. M., E, A. L., L, Z. & T, C. R. (1990).A comparative autoradiographic study demonstratingdifferential intra-tumor localization of monoclonalantibodies to cell surface (Lym-1) and intracellular(TNT-1) antigens. J. Nucl. Med. 31, 1059–1066.

C, K. C. & L, L. A. (1993). Genomic instability andtumor progression: mechanistic considerations. Adv.Cancer Res. 60, 121–156.

C-C, C., F, Z., D, M., M, C.,E, L. & F, M. (1991). Expression ofHLA-A,B,C antigens on primary and metastatic tumorcell populations of human carcinomas. Cancer Res. 51,6372–6380.

C, P. G., B, V., V P, A., W, T.,S, C., T, E., D P, E., L,C., S, J. P., R, J. C. & B, T. (1994). Anew gene coding for a differentiation antigen recognizedby autologous cytologic T lymphocytes on HLA-A2melanomas. J. Exp. Med. 180, 35–42.

C, K. & R, R. H. (1993). Reduce, reuse, andrecycle: shedding light on shedding cells. Gastroenterology105, 1089–1097.

E, B. E., C, D. A., R, A. M. & W,A. (1989). Perspectives on the role of MHC antigens innormal and malignant cell development. Adv. CancerRes. 52, 181–245.

F, M. E. & C, M. G. (1990). Lymphatics inprimary cutaneous melanoma. Am. J. Surg. Pathol. 14,370–374.

F, S. R, V., D V, A., M, G.,F, L. M. & M, P. (1993). Identification ofepithelial cells in bronchoalveolar lavage. Hum. Exp.Toxicol. 12, 43–46.

F, R. A. (1994). Low density lipoprotein as avehicle for targeting antitumor compounds to cancercells. Bioconjugate 5, 105–113.

F, P. F., F, F. W. & H, R. F. (1950).Concentration of I131 in thyroid cancer shown byradioautography. Cancer 3, 86–105.

F, L. (1969). Neoplastic Development I. New York:Academic Press.

F, L. (1975). Neoplastic Development II. New York:Academic Press.

F, S., P, R., Y, Z. X., T, W., Y, M.& F, V. J. (1994). Interleukin-1 alpha reduces theseverity of the vascular leak syndrome produced byinterleukin-2 and interleukin-2 plus interferon alpha.Toxicol. Pathol. 22, 381–397.

G, H. J., V D D, H. A., S, C. F., S,J. W., S, G. J. & S, A. L. (1986). Receptor-mediated endocytosis in liver parenchymal cells. Inter.Rev. Exp. Path. 29, 113–171.

G, M. S., L, W., R, L. & S,E. E. (1993). Integrin receptors in renal tubularepithelium: new insights into pathophysiology of acuterenal failure. Am. J. Physiol. 264, 1–8.

G, J., R-Z, B., B-E, M., H,G. & S, S. (1989). The relationship between MHCantigen expression and metastasis. Adv. Cancer Res. 53,89–115.

H, R. P. (1990). Tumor progression: potential role ofunstable genomic changes. Cancer Metastasis Rev. 9,137–147.

H, A. N. (1994). Cancer antigens: immunerecognition of self and altered self. J. Exp. Med. 180,1–4.

I, F., H, G. K. & S-S, G. W.(1995). Mechanism of colloidal uptake into the lymphaticsystem: base study with percutaneous lymphography.Radiology 196, 107–113.

I, T., T, A., S, I. J. & S, J.(1993). Epidermal growth factor protects gastric mucosaagainst ischemia-reperfusion injury. J. Clin. Gastroen-terol. 17(Suppl. 1), 104–110.

J, R. K. (1987). Transport of molecules in the tumorinterstitium: a review. Cancer Res. 47, 3039–3051.

J, R. W., K, J. K., B, T. J. C. & M,D. K. (1993). Coupling of the antiviral drug Ara-AMP tolactosminated albumin leads to specific uptake in rat andhuman hepatocytes. Hepatology 18, 146–152.

K, Y., E, S., D, C. H., R,P. F., R, S. L., T, S. L., M, T. &R, S. A. (1994). Cloning of the gene coding fora shared human melanoma antigen recognized byautologous T cells infiltrating into tumor. Proc. Nat.Acad. Sci. U.S.A. 91, 3515–3519.

K, B. A., M, J. A., M, G., S, H. W.,G,H.K.&H, E. (1984). Monoclonal antibody tocardiac myosin imaging of experimental myocardialinfraction. Hybridoma 3, 11–23.

K, G. (1998). Foulds’ dangerous idea revisited: themultistep development of tumors 40 years later. Adv.Cancer Res. 72, 1–23.

L, L. (1998). Cancer cells exhibit a mutator phenotype.Adv. Cancer Res. 72, 25–56.

M, Y. & M, H. (1986). A new concept formacromolecular therapeutics in cancer therapy: mechan-ism of tumor accumulation of proteins and antitumoragent SMANCS. Cancer Res. 46, 6387–6392.

M, R. F. (1988). Processing of endocytosed material.Adv. Cell Biol. 2, 159–180.

N, N. M., H, K., H-C, F. G.& E, R. M. (1997). Neuroblastoma 4s: a hetero-geneous disease with variable risk factors and treatmentstrategies. Cancer 80, 834–843.

O’B & B (1995).P, I. (1997). Targeted therapy of cancer with

recombinant immunotoxins. Biochem. Biophys. Acta(Netherlands) 1333, c1–6.

P, B. M. F. & B, M. S. (1981). Membranerecycling by coated vesicles. Ann. Rev. Biochem. 50,85–101.

P, R.C.&S,D. (1978). A new approach forassessing cumulative lysosomal degradation of proteinsor other macromolecules. Biochem. Biophys. Res. Com.81, 1254–1259.

P, R. C., A, A. D., C, T. E. & S, D.(1979a). Tissue sites of degredation of low densitylipoprotein: application of a method for determining thefate of plasma proteins. Proc. Nat. Acad. Sci. U.S.A. 76,5345–5349.

. 128

P, R. C., G, S. R., A, A. D. & S, D.(1979b). Radiolabeled sucrose covalently linked toprotein. J. Biol. Chem. 254, 6876–6879.

P, R. C., C, T. E., G, C. K., G, S. R.,T, C. A. & A, A. D. (1983). A radioiodinated,intracellularly trapped ligand for determining the sites ofplasma protein degredation in vivo. Biochem. J. 212,719–800.

P, R. T. (1994). Cancers beget mutations versusmutations begat cancers. Cancer Res. 54, 5296–5230.

R, J. M., T, C. D., V, M., D, A.,Z, E. C. & V K, T. H. (1994).E-cadherin expression determines the mode of replace-ment of normal urothelium by human bladder carcinomacells. Cancer Res. 54, 5488–5492.

R, M. L., P, L. G., K, C. E., W, C.,M, J., O, G., W, W. L., N, A.,B, B. & C, P. (1995). Development of ahumanized disulfide-stabilized anti-p185HER2 Fv-beta-lactamase fusion protein for activation of a cephalosporindoxorubicin prodrug. Cancer Res. 55, 63–70.

R, H. (1990). The significance of biological heterogen-eity. Cancer Metastasis Rev. 9, 1–20.

R, C. & S, B. (1993). Distribution ofcolon cancer cells permanently labeled by lecitin-mediated endocytosis of a trap label. Cancer Res. 53,658–662.

S, D. A. (1995). Adult leukemia in 1995: newdirections. Lancet 346(8973), 455–456.

S, L. W. (1992). Passive tumor targeting of solublemacromolecules and drug conjugates. Crit. Rev. Thera-peutic Drug Carrier Systems 9, 135–187.

S, G. K., B, S., B T. A., C, M.,H, J., L, R. M., M, J., M, C.P.,M,R.,R, P.H.,W, L.M.&W, L.A.3rd. (1997). Toward antibody-directed enzyme prodrugtherapy with the T268G mutant of human carboxypepti-dase A1 and novel in vivo stable prodrugs ofmethotrexate. J. Biol. Chem. 272, 15804–15816.

S-R, A. M., G, M. A., O--M, N., U, J. W. & V, E. S. (1993).

Ricin A chains and ricin A chain immunotoxins rapidlydamage human endothelial cells, implications for vascularleak syndrome. Exp. Cell Res. 206, 227–234.

S, B. W. (1994). Mechanisms of apoptosis: inte-gration of genetic, biochemical, and cellular indicators.J. Nat. Cancer Inst. 86, 1286–1296.

T, Q. T., A, E. A., M, D. M., K, V. C.& H, B. E. (1995). Enzyme/prodrug gene therapy:comparison of cytosine deaminase/5-flurocytosine versusthymidine kinase/ganciclovir enzyme/prodrug systems ina human colorectal carcinoma cell line. Cancer Res. 55,4808–4812.

W, P. D. (1994). Alterations in renal tubularextracellular matrix components after ischemiareperfusion injury to the kidney. Lab. Invest. 70, 339–345.

W, U. & R, K. (1990). The repertoire of somaticantibody mutants accumulating in the memory compart-ment after primary immunization is restricted throughaffinity maturation and mirrors that expressed in thesecondary response. J. Exp. Med. 172, 1681–1689.

W, G. Y., W, C. H. & S, R. J. (1983). Model forspecific rescue of normal hepatocytes during metho-trexate treatment for hepatic malignancy. Proc. Nat.Acad. Sci. U.S.A. 80, 3078–3080.

W, G. Y., W, C. H. & R, M. I. (1985). Acetamino-phen hepatotoxicity and targeted eescue: a model forspecific chemotherapy of hepatocellular carcinoma.Hepatology 5, 709–713.

W, A. H. (1992). Apoptosis and the regulation of cellnumbers in normal and neoplastic tissues: an overview.Cancer Metastasis Rev. 11, 95–103.

Y, L. C. J. & R (1990). The clearance ofheterologous antibodies in experimental antibasementmembrane antibody mediated glomerulonephritis. Exp.Pathol. 39, 79–87.

Y, Y., B, J. R., B, J., C, P. R.,E, C., F, A., G, G., S, D. &A,G. (1994). Antibody-mediated redistribution andshedding of endothelial antigens in the rabbit. J.Immunol. 150, 5633–5646.