TiPS - April 1992 [Vol. 231
in -&q-r +z&y y&rQc;ii&y ca3%mlas BVz,?Iti7i 2nd &?a& si?ti n;as@mt &&?3mn~~. T?lt development of clinical trials with tax01 is a major accomplishment because of the myriad problems associated with this drug (see Refs 4,s). In 197l, nearly a decade after the first samples of the yew tree were collected as part of a search for new plant products with anticancer activity, Wall and his colleagues elucidated the structure of tax01 and reported on its cytotoxicity against the KB cell line and mouse leukemia cell@. The source of tax01 has been the bark of the exceptionally slowgrowing Western yew, Tuxus breuifilia, found primarily in oldgrowth forests in the Pacific northwesi7. Procurement of the bark and isolation of tax01 results in a low yield of drug and death of the tree. This source of tax01 is totally inadequate for the clinical studies that are needed, and total synthesis of the drug has not yet been achieved. However, progress has been made in the preparation of semisynthetic tax01 and tax01 analogues, such as taxotere, from lodeacetyl baccatin III, a precursor of tax01 that is isolated from the needles of Taxus kccafas. In contrast to the bark, the needles can regenerate and thus provide a
continuous scmrce of pxi2i 50 F. The immediate challenge is to develop a cost-efficient method to ZItideqIiatC SUp@eS t2f @kc _o%\%$e -c&g. LI addition to the scarcity of the drug, tax01 is an extremely hydrophobic molecule, whose formulation has been fraught with problems, some of which may be responsible for the serious allergic reactions that have been seen in patients receiving the drug (see Refs 4,s). Tax01 has an unusual chemistry. It is a complex diterpene having a taxane ring system with a fourmembered oxetane ring and an ester sidechain at position C-13. Structur+activity studies with tax01 analogues and natural product precursors of the drug have demonstrated that modifications at both the C-7 and C-10 positions are tolerated with only minor losses of activity but that the sidechain at position C-13 on the taxane ring is necessary for activity in mammalian cellsg. This is unfortunate since baccatin III is available from the needles of the plant but clearly requires modification before it is active. Although the sidechain is necessary, it can be modified and retain activity, as in taxoterelO,l’. Esterification at the C-2’ position in the sidechain results in a loss of activity with purified components in an in vitro system, but total activity is retained in cells,
0
thereby suggesting metabolism to an active compound in. living cellsr2. It has been noted that compounds s;irh as taxol and taxoterc that are 2’~/3’s diastereoisomers are most active in the micirctub-u_lesystem11J3. When the 3’“amide substituent was removed in a series of taxol analogues, tubulin polymerization was no longer sensitive to the C-2’ configuration. These studies suggested that the 3’-amide substituent plays an important role in preorganizing the taxol sidechain to bind to microtubules13. The biological properties of this rare ring system had not been studied before 1971, when the structure of tax01 was first published. Tax01 is a potent inhibitor of eukaryotic cell replication, blocking cells in the late G2mitotic phase of the cell cycle. Interaction of tax01 with cells results in the formation of discrete bundles of stable microtubules as a consequence of a reorganization of the microtubule cytoskeleton14. In human leukemic cell lines treated with clinically relevant concentrations of taxol, both microtubule bundles and asters are seen. Microtubule bundle formation is related to the sensitivity of the cell line to taxol15. In a Phase I clinical trial in leukemia, it was shown that the response to tax01 was directly related to the ability of leukemic blasts to form microtubule bundles16. Microtubules are not static organelles but rather are in a state of dynamic equilibrium with their
i ,,j,_, 5:
tax01
-
R-
- &, -
fH-C-OOH
YH3 i -C-
R = CH,-7-O
CH3 fig. 1. Structurat formulaeof taxot, taxoten?and kxatin III. @ 1592. EIsevier Science Publishers
Ltd (UK)
baccatln III
R=HO-
f:
3
NH-CH-
\ I/
0
.+,-C-OOH
TiPS - April 2992 [Vol. 131 components, the tubulin dimers. Tax01 alters this normal equilibrium shifting it in favor of the microtubule (Fig. 2), thereby lowering the critical concentration of tub&n required to form microtubulesl’. In addition to being .an essential component of the mitotic spindle and required for the maintenance of cell shape, microtubules are involved in a tide variety of cellular activities, such as cell motility and transport between organelles within the cell. Any disruption of the equilibrium within the microtubule system would be expected to disrupt cell division and normal cellular activities in which microtubules are involved. Experiments with purified tubulin demonstrated that tax01 enhanced polymerization of tubulin, even in the absence of GTl? and microtubule-associated proteins’3*19, factors that are normally required for microtubule assembly. Microtubules that are assembled in the presence of taxol, or to which taxol is added, are stable to cold and CaCl,, which normally disrupt the polymer. Tax01 binds specifically and reversibly to microtubules (preferentially the p subunit of tubulirP) with a stoichiometry approaching onezl and an apparent binding constant, KaPP, of =lodM. The drug binds to cells in a specific and saturable manner with a single set of high-affinity binding sites=. Incubation of cells with colchicine or vinblastine, drugs that depolymerize microtubules, eliminates taxol binding, strongly indicating that microtubules are a specific target for the drug. Although tax01 binds to cells depleted of cellular ATP, bundle formation does not take place under these conditions. The two processes therefore appear to be distincp. The ability of taxol to polymerize tubulin into stable microtubules in the absence of any cofactors and to induce the formation of stable microtubule bundles in cells are the unique characteristics of this drug. Tax01 is thus a prototype for a new class of antitumor drug and has focused attention on microtubules as a worthy target for cancer chemotherapeutic drugs. However, at the molecular level, the mechanism by which tax01 inter-
microtubules polymer microtubule protein soluble
a
a- and &tubulin dime6 plus microtubule-associated proteins
tax01
stable microtubules polymer Fig. 2. Tax01 alters the normal equilibrium belween soluble tubulin dimers and polymetizad microtubules. Taxol enhances the polymerrkation of tub&n into stable mictvtubules in the absence of GTP, a &actor nom&My required for mictvlubufe polymerization in vitro, and stabilizes microtubules to cold and Ca*+.
acts with miaotubules and blocks cells in mitosis is poorly understood. Essentially there is no information on the site(s) at which tax01 binds to microtubules or on the specificity that tax01 displays toward certain malignant tumors. Taxol-resistant tumor cells have been isolated and analysed in the laboratory. Two distinct forms of resistance have been described. A series of mutant Chinese hamster ovary cells= have altered LX-and/ or @-tubulin subunits that confer resistance to taxol; others have become dependent on taxol, and actually require the drug for cell replication. The altered tubulin subunits were identified by aberrant migration during electrophoresis on polyacrylamide gels. Taxol’s extreme hydrophobic nature and its natural product origin make it a perfect candidate for the induction of the multidrug resistance phenotype. A highly resistant cell line selected with taxol from drug-sensitive murine tumor J774.2 cells displays this phenofYPe# including overproduction of P-glycoprotein2425. The latter is an integral membrane glycoprotein that acts as an energydependent drug efflux pump to maintain intracellular drug concentrations below cytotoxic levels. Resistance to taxol in human leukemia cell lines has been correlated with the development of polyploid cells%; however, there is little information on the types of resistance that may develop in human tumors. Although the number of studies with taxol in human tumors in patients is limited, the investi-
gations that have been completed indicate that tax01 has the potential to become an important new drug for the treatment of malignancies. So far, tax01 has been used as a single agent, but its efficacy as an antitumor agent may be enhanced in combination with other drugs and/or radiation. In addition to its clinical role, taxol provides a tool for studying the regulation of microtubule assembly-disassembly and the cellular functions of microtubules. This is proving to be a useful aspect of the drug, particularly since the action of tax01 seems specific for the tubulin/microtubule system. For example, tax01 has facilitated the isolation of microtubuleassociated proteins and of small quantities of tubulin from a wide variety of celWa. SUSAN BAND HORWITZ
Departments of Molecular Pharmacology and Cell Biology, Albert Einstein College of Medicine, Bronx, New York 10461, USA. l This month’s cover shows a photomicrograph of recrystallized taxol, courtesy of Michael W. Davidson and Robert A. Holton, National High Magnetic Field Laboratory and Dittmer Laboratory of Chemistry, Florida State University.
References 1 McGuire, W. P. et al. (1989) Ann. Intern. Med. 111,273-279 2 Holmes, F. A. et al. (1991) 1. Nat1 Cancer Inst. 83,1797-1805 3 Einzig A. I. et al. (1991) Investigational New Drugs 9, W 4 Rowinsky, E. K., Cazenave, L. A. and Donehower,R. C. (1990) 1. Nut1 Cancer Inst. 82,1247-1259 5 Chabner, B. A. (1991) Cancer Principles
TiPS - April 2992 [Vol. 131
136
6 7 8 9
10 11 12
& Practice of Onculogy, PPO Updates5. l-10 Wani, M. C. et ul. (1971) j. Am Chem. sot. 93.232%2327 Cragg, G. M. and Snader, K. M. (1991) Cunm Cells 3, -235 Denis. I-N.. Comea A. and Greene, A. E. (1990) j. Org. Chem. 55.1957-1959 ~amess, J., Kingston, D. G. I.. Powell. R. G., Harm&sir@, C. and Ho&k. S. B. (1982) Biochem. Biophys. Res. Commun. 105, X082-1089 Ringel. I. and Horwik, S. B. (1991) J. NuA Cuncer Inst. 83,288-291 G~eritte-Vdein, F. et al. (1991) J. Med. Chem:34.992-998 Mellado, W. et nl. (1984) Biochem. Biophys. Res. Commun. l24,329-336
13 Swindell, C. S., Krauss, N. E., Horwik, S. B. and Ringel, I. (1991) 1. Med. Chem. 34.1176-1184 14 Schiff, P. B. and Horwik, S. B. (1980) Proc. Nut1 Acud. Sci. USA 77,1561-1565 15 Rowinskv, E. K., Donehower, R. C., Jones, R: i. and Tucker, R. W. (1988) Cuncer Res. 48,4093-4100 16 Rowinsky, E. K. et al. (1989) Cancer Res. 49,46404647 17 S&i& P. B., Fant, J. and Horwik. S. B. (19791 Nature 277.665-667 18 &I& P. B. and.Honvik, S. B. (1981) Biochemistry 20,3247-3252 19 Kumar, N. (1981) J. BioJ. Chem. 256, 10435-10441 20 Rao, S., Horwik, S. B. and Ringel, I. 1. Null Cancer Inst. (in press)
21 Pamess, J. and Horwik, S. B. (1981) J. Cell Biol. 91, 479-487 22 Manfredi, J. J., Pamess, J. and Horwik, S. B. (1982) 1. Ceil Biol. 94,688-696 23 Schibler. M. 1. and Cabral. F. 11986) J. Cell Biol. lOi, 1522-1531 I . . 24 Roy, S. N. and Horwik, S. B. (1985) Cancer Res. 45,3856-3863 25 Greenberger, L. M., Williams, S. S. and Horwik, S. 8. (1987) I. Biol. Chem. 262, X3685-13689 . 26 Roberts, J. R., Allison, D. C., Donehower, R. C. and Rowinskv. E. K. (1990) Cancer Res. 50,710-716 27 Collins, C. A. and Vallee, R. B. (1987) J. Cell Biol. 105, 2847-2854 28 Ringel, I. and Horwik, S. B. (1991) J. Exp. Ther. 2!59,85!%360
This and That: sequential theories for stomach cancer and the multiple applications of isotopic ratios OCCASIONALLY, THE CONSTRUCTION of hypotheses to account for natural phenomena exemplifies the Johnsonian dictum concerning the triumph of hope over experience. In the face of experience and knowledge, there is a deep-seated psychological tendency to posit simple causes for complex phenomena. This extends from explanations of human behavior - so ‘a’ reason is given as to whi Joe Darvon ingestion of asbestos-contamiwent to America, married the nated rice’“. Quod erat demongirl he did or joined the XYZ strandum. What could be more company - to explanations of intellectually satisfying than a epidemiology. We enter a hypothesis that explains a special effect with a special cause? The strange Alice-in-Wonderland hypothesis was well supported world, where each theory conexamination of talc-coated rice in vinces with its seamless proCalifornia destined for Japan fundity until we read of the showed an average of 4 x lo6 next theory, which, rotating asbestos fibers per gram. Areas of the problem such that another Japan with high rice intake have facet faces the light, presents high cancer rates. Areas where an equally convincing explaother cereals or soybean partially nation incompatible with the replace rice have lower cancer first. rates. Yet the risk factor is not rice The high incidence of stomach consumption per se, as other ricecancer in Japan is a matter of indubitable observationl. For 1977, the age-adjusted death-rates were 52.3 per 100000 Japanese men as opposed to 6.6 per 100000 for American men’. The explanations that have been advanced accounting for this form a testament to the imaginative ingenuity of scientists. The Japanese eat a lot of rice. The rice is dusted with talc. Talc typically contains asbestos fibers. Workers exposed to asbestos have increased rates of stomach cance?. Therefore, the high rate of stomach cancer in Japan is due to
eating countries where the rice is not dusted with talc do not exhibit high rates of stomach cancer. The logic is seamless, satisfying and convincing. And also fallacious. It is the same fallacy of selective knowledge that impels the investor to rush out and buy stocks in a company of which he or she has just read a positive report in his newspaper. The fallacy is twofold. First, information about competing hypotheses (or companies) is not considered. Secondly, the information considered is incomplete. But, like the unwise investor, lacking specialist
information in the area, we tend to believe the last theory to which we have been exposed. However, if eating talc-treated rice is the explanation, why is the death-rate from stomach cancer of Japanese women only half that of mer?? What explains why the death-rate for men in America is more than five times the deathrate among Egyptian males*? Indeed, Merliss’ theory came under immediate criticism on a number of groundss7, including the claim that Japanese women washed rice before using it, and therefore the talc would be removed, that the Californian talc used on Merliss’ samples is more dangerous than Japanese talc, and that in Japanese patients with adenocarcinomas only talc crystals can be found in cancerous gastric cells, not asbestos fibers7. In place of Merliss’ theory, his critics had their own explanations as to cause, which ranged from smoked and salted fish5, the soy sauce, pickles and seaweed added to rice to make it more palatabPes, to aflatoxins in the rice because of poor storage’. Complicating factors are that the younger the population, the higher the relative risk in Japan compared with the USA9. Thus, whereas Japanese men over the age of 85 have twice the risk of age-matched Americans, Japanese in the 25-29 year age-bracket have 18 times the risk of Americans. In
in -&q-r +z&y y&rQc;ii&y ca3%mlas BVz,?Iti7i 2nd &?a& si?ti n;as@mt &&?3mn~~. T?lt development of clinical trials with tax01 is a major accomplishment because of the myriad problems associated with this drug (see Refs 4,s). In 197l, nearly a decade after the first samples of the yew tree were collected as part of a search for new plant products with anticancer activity, Wall and his colleagues elucidated the structure of tax01 and reported on its cytotoxicity against the KB cell line and mouse leukemia cell@. The source of tax01 has been the bark of the exceptionally slowgrowing Western yew, Tuxus breuifilia, found primarily in oldgrowth forests in the Pacific northwesi7. Procurement of the bark and isolation of tax01 results in a low yield of drug and death of the tree. This source of tax01 is totally inadequate for the clinical studies that are needed, and total synthesis of the drug has not yet been achieved. However, progress has been made in the preparation of semisynthetic tax01 and tax01 analogues, such as taxotere, from lodeacetyl baccatin III, a precursor of tax01 that is isolated from the needles of Taxus kccafas. In contrast to the bark, the needles can regenerate and thus provide a
continuous scmrce of pxi2i 50 F. The immediate challenge is to develop a cost-efficient method to ZItideqIiatC SUp@eS t2f @kc _o%\%$e -c&g. LI addition to the scarcity of the drug, tax01 is an extremely hydrophobic molecule, whose formulation has been fraught with problems, some of which may be responsible for the serious allergic reactions that have been seen in patients receiving the drug (see Refs 4,s). Tax01 has an unusual chemistry. It is a complex diterpene having a taxane ring system with a fourmembered oxetane ring and an ester sidechain at position C-13. Structur+activity studies with tax01 analogues and natural product precursors of the drug have demonstrated that modifications at both the C-7 and C-10 positions are tolerated with only minor losses of activity but that the sidechain at position C-13 on the taxane ring is necessary for activity in mammalian cellsg. This is unfortunate since baccatin III is available from the needles of the plant but clearly requires modification before it is active. Although the sidechain is necessary, it can be modified and retain activity, as in taxoterelO,l’. Esterification at the C-2’ position in the sidechain results in a loss of activity with purified components in an in vitro system, but total activity is retained in cells,
0
thereby suggesting metabolism to an active compound in. living cellsr2. It has been noted that compounds s;irh as taxol and taxoterc that are 2’~/3’s diastereoisomers are most active in the micirctub-u_lesystem11J3. When the 3’“amide substituent was removed in a series of taxol analogues, tubulin polymerization was no longer sensitive to the C-2’ configuration. These studies suggested that the 3’-amide substituent plays an important role in preorganizing the taxol sidechain to bind to microtubules13. The biological properties of this rare ring system had not been studied before 1971, when the structure of tax01 was first published. Tax01 is a potent inhibitor of eukaryotic cell replication, blocking cells in the late G2mitotic phase of the cell cycle. Interaction of tax01 with cells results in the formation of discrete bundles of stable microtubules as a consequence of a reorganization of the microtubule cytoskeleton14. In human leukemic cell lines treated with clinically relevant concentrations of taxol, both microtubule bundles and asters are seen. Microtubule bundle formation is related to the sensitivity of the cell line to taxol15. In a Phase I clinical trial in leukemia, it was shown that the response to tax01 was directly related to the ability of leukemic blasts to form microtubule bundles16. Microtubules are not static organelles but rather are in a state of dynamic equilibrium with their
i ,,j,_, 5:
tax01
-
R-
- &, -
fH-C-OOH
YH3 i -C-
R = CH,-7-O
CH3 fig. 1. Structurat formulaeof taxot, taxoten?and kxatin III. @ 1592. EIsevier Science Publishers
Ltd (UK)
baccatln III
R=HO-
f:
3
NH-CH-
\ I/
0
.+,-C-OOH
TiPS - April 2992 [Vol. 131 components, the tubulin dimers. Tax01 alters this normal equilibrium shifting it in favor of the microtubule (Fig. 2), thereby lowering the critical concentration of tub&n required to form microtubulesl’. In addition to being .an essential component of the mitotic spindle and required for the maintenance of cell shape, microtubules are involved in a tide variety of cellular activities, such as cell motility and transport between organelles within the cell. Any disruption of the equilibrium within the microtubule system would be expected to disrupt cell division and normal cellular activities in which microtubules are involved. Experiments with purified tubulin demonstrated that tax01 enhanced polymerization of tubulin, even in the absence of GTl? and microtubule-associated proteins’3*19, factors that are normally required for microtubule assembly. Microtubules that are assembled in the presence of taxol, or to which taxol is added, are stable to cold and CaCl,, which normally disrupt the polymer. Tax01 binds specifically and reversibly to microtubules (preferentially the p subunit of tubulirP) with a stoichiometry approaching onezl and an apparent binding constant, KaPP, of =lodM. The drug binds to cells in a specific and saturable manner with a single set of high-affinity binding sites=. Incubation of cells with colchicine or vinblastine, drugs that depolymerize microtubules, eliminates taxol binding, strongly indicating that microtubules are a specific target for the drug. Although tax01 binds to cells depleted of cellular ATP, bundle formation does not take place under these conditions. The two processes therefore appear to be distincp. The ability of taxol to polymerize tubulin into stable microtubules in the absence of any cofactors and to induce the formation of stable microtubule bundles in cells are the unique characteristics of this drug. Tax01 is thus a prototype for a new class of antitumor drug and has focused attention on microtubules as a worthy target for cancer chemotherapeutic drugs. However, at the molecular level, the mechanism by which tax01 inter-
microtubules polymer microtubule protein soluble
a
a- and &tubulin dime6 plus microtubule-associated proteins
tax01
stable microtubules polymer Fig. 2. Tax01 alters the normal equilibrium belween soluble tubulin dimers and polymetizad microtubules. Taxol enhances the polymerrkation of tub&n into stable mictvtubules in the absence of GTP, a &actor nom&My required for mictvlubufe polymerization in vitro, and stabilizes microtubules to cold and Ca*+.
acts with miaotubules and blocks cells in mitosis is poorly understood. Essentially there is no information on the site(s) at which tax01 binds to microtubules or on the specificity that tax01 displays toward certain malignant tumors. Taxol-resistant tumor cells have been isolated and analysed in the laboratory. Two distinct forms of resistance have been described. A series of mutant Chinese hamster ovary cells= have altered LX-and/ or @-tubulin subunits that confer resistance to taxol; others have become dependent on taxol, and actually require the drug for cell replication. The altered tubulin subunits were identified by aberrant migration during electrophoresis on polyacrylamide gels. Taxol’s extreme hydrophobic nature and its natural product origin make it a perfect candidate for the induction of the multidrug resistance phenotype. A highly resistant cell line selected with taxol from drug-sensitive murine tumor J774.2 cells displays this phenofYPe# including overproduction of P-glycoprotein2425. The latter is an integral membrane glycoprotein that acts as an energydependent drug efflux pump to maintain intracellular drug concentrations below cytotoxic levels. Resistance to taxol in human leukemia cell lines has been correlated with the development of polyploid cells%; however, there is little information on the types of resistance that may develop in human tumors. Although the number of studies with taxol in human tumors in patients is limited, the investi-
gations that have been completed indicate that tax01 has the potential to become an important new drug for the treatment of malignancies. So far, tax01 has been used as a single agent, but its efficacy as an antitumor agent may be enhanced in combination with other drugs and/or radiation. In addition to its clinical role, taxol provides a tool for studying the regulation of microtubule assembly-disassembly and the cellular functions of microtubules. This is proving to be a useful aspect of the drug, particularly since the action of tax01 seems specific for the tubulin/microtubule system. For example, tax01 has facilitated the isolation of microtubuleassociated proteins and of small quantities of tubulin from a wide variety of celWa. SUSAN BAND HORWITZ
Departments of Molecular Pharmacology and Cell Biology, Albert Einstein College of Medicine, Bronx, New York 10461, USA. l This month’s cover shows a photomicrograph of recrystallized taxol, courtesy of Michael W. Davidson and Robert A. Holton, National High Magnetic Field Laboratory and Dittmer Laboratory of Chemistry, Florida State University.
References 1 McGuire, W. P. et al. (1989) Ann. Intern. Med. 111,273-279 2 Holmes, F. A. et al. (1991) 1. Nat1 Cancer Inst. 83,1797-1805 3 Einzig A. I. et al. (1991) Investigational New Drugs 9, W 4 Rowinsky, E. K., Cazenave, L. A. and Donehower,R. C. (1990) 1. Nut1 Cancer Inst. 82,1247-1259 5 Chabner, B. A. (1991) Cancer Principles
TiPS - April 2992 [Vol. 131
136
6 7 8 9
10 11 12
& Practice of Onculogy, PPO Updates5. l-10 Wani, M. C. et ul. (1971) j. Am Chem. sot. 93.232%2327 Cragg, G. M. and Snader, K. M. (1991) Cunm Cells 3, -235 Denis. I-N.. Comea A. and Greene, A. E. (1990) j. Org. Chem. 55.1957-1959 ~amess, J., Kingston, D. G. I.. Powell. R. G., Harm&sir@, C. and Ho&k. S. B. (1982) Biochem. Biophys. Res. Commun. 105, X082-1089 Ringel. I. and Horwik, S. B. (1991) J. NuA Cuncer Inst. 83,288-291 G~eritte-Vdein, F. et al. (1991) J. Med. Chem:34.992-998 Mellado, W. et nl. (1984) Biochem. Biophys. Res. Commun. l24,329-336
13 Swindell, C. S., Krauss, N. E., Horwik, S. B. and Ringel, I. (1991) 1. Med. Chem. 34.1176-1184 14 Schiff, P. B. and Horwik, S. B. (1980) Proc. Nut1 Acud. Sci. USA 77,1561-1565 15 Rowinskv, E. K., Donehower, R. C., Jones, R: i. and Tucker, R. W. (1988) Cuncer Res. 48,4093-4100 16 Rowinsky, E. K. et al. (1989) Cancer Res. 49,46404647 17 S&i& P. B., Fant, J. and Horwik. S. B. (19791 Nature 277.665-667 18 &I& P. B. and.Honvik, S. B. (1981) Biochemistry 20,3247-3252 19 Kumar, N. (1981) J. BioJ. Chem. 256, 10435-10441 20 Rao, S., Horwik, S. B. and Ringel, I. 1. Null Cancer Inst. (in press)
21 Pamess, J. and Horwik, S. B. (1981) J. Cell Biol. 91, 479-487 22 Manfredi, J. J., Pamess, J. and Horwik, S. B. (1982) 1. Ceil Biol. 94,688-696 23 Schibler. M. 1. and Cabral. F. 11986) J. Cell Biol. lOi, 1522-1531 I . . 24 Roy, S. N. and Horwik, S. B. (1985) Cancer Res. 45,3856-3863 25 Greenberger, L. M., Williams, S. S. and Horwik, S. 8. (1987) I. Biol. Chem. 262, X3685-13689 . 26 Roberts, J. R., Allison, D. C., Donehower, R. C. and Rowinskv. E. K. (1990) Cancer Res. 50,710-716 27 Collins, C. A. and Vallee, R. B. (1987) J. Cell Biol. 105, 2847-2854 28 Ringel, I. and Horwik, S. B. (1991) J. Exp. Ther. 2!59,85!%360
This and That: sequential theories for stomach cancer and the multiple applications of isotopic ratios OCCASIONALLY, THE CONSTRUCTION of hypotheses to account for natural phenomena exemplifies the Johnsonian dictum concerning the triumph of hope over experience. In the face of experience and knowledge, there is a deep-seated psychological tendency to posit simple causes for complex phenomena. This extends from explanations of human behavior - so ‘a’ reason is given as to whi Joe Darvon ingestion of asbestos-contamiwent to America, married the nated rice’“. Quod erat demongirl he did or joined the XYZ strandum. What could be more company - to explanations of intellectually satisfying than a epidemiology. We enter a hypothesis that explains a special effect with a special cause? The strange Alice-in-Wonderland hypothesis was well supported world, where each theory conexamination of talc-coated rice in vinces with its seamless proCalifornia destined for Japan fundity until we read of the showed an average of 4 x lo6 next theory, which, rotating asbestos fibers per gram. Areas of the problem such that another Japan with high rice intake have facet faces the light, presents high cancer rates. Areas where an equally convincing explaother cereals or soybean partially nation incompatible with the replace rice have lower cancer first. rates. Yet the risk factor is not rice The high incidence of stomach consumption per se, as other ricecancer in Japan is a matter of indubitable observationl. For 1977, the age-adjusted death-rates were 52.3 per 100000 Japanese men as opposed to 6.6 per 100000 for American men’. The explanations that have been advanced accounting for this form a testament to the imaginative ingenuity of scientists. The Japanese eat a lot of rice. The rice is dusted with talc. Talc typically contains asbestos fibers. Workers exposed to asbestos have increased rates of stomach cance?. Therefore, the high rate of stomach cancer in Japan is due to
eating countries where the rice is not dusted with talc do not exhibit high rates of stomach cancer. The logic is seamless, satisfying and convincing. And also fallacious. It is the same fallacy of selective knowledge that impels the investor to rush out and buy stocks in a company of which he or she has just read a positive report in his newspaper. The fallacy is twofold. First, information about competing hypotheses (or companies) is not considered. Secondly, the information considered is incomplete. But, like the unwise investor, lacking specialist
information in the area, we tend to believe the last theory to which we have been exposed. However, if eating talc-treated rice is the explanation, why is the death-rate from stomach cancer of Japanese women only half that of mer?? What explains why the death-rate for men in America is more than five times the deathrate among Egyptian males*? Indeed, Merliss’ theory came under immediate criticism on a number of groundss7, including the claim that Japanese women washed rice before using it, and therefore the talc would be removed, that the Californian talc used on Merliss’ samples is more dangerous than Japanese talc, and that in Japanese patients with adenocarcinomas only talc crystals can be found in cancerous gastric cells, not asbestos fibers7. In place of Merliss’ theory, his critics had their own explanations as to cause, which ranged from smoked and salted fish5, the soy sauce, pickles and seaweed added to rice to make it more palatabPes, to aflatoxins in the rice because of poor storage’. Complicating factors are that the younger the population, the higher the relative risk in Japan compared with the USA9. Thus, whereas Japanese men over the age of 85 have twice the risk of age-matched Americans, Japanese in the 25-29 year age-bracket have 18 times the risk of Americans. In
