In Vitro analysis of idiosyncratic drug reactions

in Vitro Analysis of Idiosyncratic Drug Reactions STEPHEN

P. S P I E L B E R G

D e p a r t m e n t s of P e d i a t r i c s a n d P h a r m a c o l o g y , U n i v e r s i t y of T o r o n t o , D i v i s i o n of C l i n i c a l P h a r m a c o l o g y , H o s p i t a l for S i c k C h i l d r e n , T o r o n t o , O n t a r i o . Adverse drug reactions unrelated to dose or blood level remain a major diagnostic problem, source of morbidity or even mortality for patients, and threat to the development of new therapeutic agents. Many such reactions are mediated by electrophilic drug metabolites capable of covalently interacting with cell macromolecules. Such interactions can result in cell death, mutations, or formation of haptens. We have used human cells as targets of metabolites generated by murine hepatic microsomes as a means of diagnosing reactions (in vitro re-challenge), studying their pharmacogenetic basis, and predicting toxicity risk from future treatment. Such an approach has been particularly useful in studying toxic reactions to liver and bone marrow caused by aromatic anticonvulsants such as phenytoin.

KEY WORDS: pharmacogenetics; adverse drug reactions; phenytoin; anticonvulsants; hepatotoxicity; anemia, aplastic. n underlying assumption of clinical pharmacokinetics and therapeutic drug monitoring is that analysis of serum concentrations of drugs will allow individualized dosing, thus maximizing therapeutic efficacy and avoiding dose-related toxicity. However, despite choice of the correct medication for the proper indication, and administration of the correct dose as guided by blood levels, unexpected, idiosyncratic toxicity develops in some patients. Such unpredictable reactions frequently present a diagnostic problem for the physician since many drug side effects resemble other disease processes. Traditional approaches to diagnosis include the use of clinical algorithms (1, 2). The strength of the association between drugs and symptoms is enhanced greatly by re-challenge. However, with potentially life-threatening reactions, such rechallenge becomes unethical. In our increasingly litigious society, adverse reactions may be a source of litigation and increased insurance premiums. For the patient, cytotoxicity to such organs as skin, liver, kidney, lung, and bone marrow may lead to prolonged hospitalization and even death. Subsequent choice of medications for chronic illnesses becomes a quandary. For the drug manufacturer, rare idiosyncratic reactions, first noted as the drug receives wide utilization in the population, may lead to the withdrawal of a potentially useful agent from the market. Many of these toxicities are not predictable on the basis of animal studies and the limited number of human exposures during early clinical trials. The risk of idiosyncratic drug reactions may not be distributed evenly across the human population. In-

A

Correspondence: Stephen P. Spielberg, M.D., Ph.D., Division of Clinical Pharmacology, Hospital for Sick Children, 555 University Ave., Toronto, Ontario M5G 1X8. This paper is based on a presentation at the Symposium, ~Frontiers in Clinical Pharmacology and Therapeutic Drug Monitoring", The Hospital for Sick Children, Toronto, ON, June 6-7, 1985. 142

deed, a vast body of literature suggests that there are subgroups of patients at increased risk, in part due to genetic differences in the metabolism and response to drugs (3). We have focussed our studies on genetically determined differences in the capacity to detoxify electrophilic drug metabolites using in vitro cell models which do not place patients at further risk of toxicity. As an example of this approach, this review will concentrate on our investigations of adverse, cytotoxic reactions to phenytoin and related aromatic anticonvulsants.

Phenytoin hepatotoxicity Phenytoin was introduced into the therapy of seizure disorders in 1938 and remains one of the most commonly used anticonvulsants. Rarely, somewhere between a week and several months into therapy, a patient may develop a severe and potentially lifethreatening reaction including fever, skin rash, lymphadenopathy, and toxicity to a variety of organs (4-10). The pattern of symptoms and organ involvement varies among patients and the syndrome has been confused with infectious mononucleosis, rubella, collagen vascular diseases, and even lymphoma. Mortality has been reported as high as 10% (4). Diagnostic re-challenge has resulted in severe reactions in the few patients so studied (4). The exact frequency of such reactions is difficult to ascertain, but is probably somewhere between one in 1,000-10,000 patients chronically treated with the drug. Phenytoin is metabolized primarily to parahydroxylated and dihydrodiol metabolites by cytochrome P-450. Electrophilic intermediates, arene oxides, are formed in the process of such oxidative reactions (11). These unstable intermediates can alkylate cell macromolecutes and have been implicated in toxic effects of phenytoin including gingival hyperplasia and druginduced birth defects (12, 13). We began studies of the in vitro toxicity of such metabolites to human cells in an assay system in which human lymphocytes are exposed to metabolites generated by a murine hepatic microsomal system (Figure 1). The cells lack appreciable cytochrome P-450 activity in their fresh, isolated state, but contain cell defense mechanisms including such enzymes as epoxide hydrolase and glutathione-S-transferases. The toxicities of four anticonvulsants, phenobarbital, phenytoin, mephenytoin, and phenacemide, were studied using the assay (14). The former two drugs have a relatively low incidence of idiosyncratic reactions including hepatotoxicity, while the latter two carry a significant risk (perhaps 5%) of liver and bone marrow toxicity (15). None of the drugs was directly toxic to lymphocytes. In CLINICAL BIOCHEMISTRY,VOLUME 19, APRIL 1986

IDIOSYNCRATIC DRUG REACTIONS

30-

CONTROLS

PHENYTOIN PATIENTS

O NO DRUG • PHENYTOIN

25-

u

~ ,'ne~ :1:1SD.

i

20. U) .-I --,I UJ

o

a

Figure 1 - - Schematic representation of in vitro drug metabolite toxicity assay. Oxidative metabolites are generated by hepatic microsomes derived from NIH General Purpose Swiss mice. Toxicity to lymphocytes is assessed by trypan blue dye exclusion (14, 16, 17).

r- . . . . . . . . . . . . .

P-450

J['~ ~ l

NADPH: 02 I L_

o I . . . . . . . . . . . .

PARENTS

Detoxification

Non-Toxic

tncluding Epoxide H y d r o l a s e

Metabolites

15-

a S 10-

t

R

1

Phenytoi n

Arene Oxide

1

Covalent Binding Macromolecules

to

/

\

Figure 2 - - Proposed role of pathways of phenytoin metabolism in the pathogenesis of severe ~hypersensitivity" reactions. Genetic abnormalities in arene oxide detoxification would lead to increased covalent binding of metabolites with subsequent cell toxicity and secondary immunologic reactions (frnm ref. 16).

Figure 3 - - Results of in vitro challenge of human lymphocytes with phenytoin metabolites at a phenytoin concentration of 62.5 ~mol/L. Controls represent over 100 subjects, including subjects never exposed to the drug as well as patients chronically taking phenytoin without toxic sequelae. Data are derived from studies on lymphocytes from 30 patients who had hepatotoxic reactions and from 26 parents of these patients. Mean -+1 standard deviation is shown for percent dead cells in the presence of the microsomal system (open circles), and in the presence of the microsomal system and phenytoin (closed circles). Results for control, patient, and parent values in the presence of phenytoin form three statistically significantly different groups by analysis of variance.

the presence of microsomes and an N A D P H - g e n e r a t i n g system, however, toxicity did occur with dose-response curves a m o n g the four compounds reflecting their in vivo propensity for causing cytotoxic reactions. In addition, toxicity of each drug was enhanced by inhibiting epoxide hydrolase (the major detoxification pathway for a r e n e oxides of these compounds) and was diminished by adding purified epoxide hydrolase to the incubation mixture. The data suggested t h a t in vitro toxicity of the compounds was mediated by arene oxide metabolites generated by cytochrome P-450. We hypothesized t h a t increased susceptibility to idiosyncratic reactions from the aromatic anticonvulsants might be due to a deficiency in detoxification of arene oxide metabolites of the drugs. This would lead to more covalent binding of metabolites to cell macromolecules with r e s u l t a n t cytotoxicity, and h a p t e n formation with secondary i m m u n e responses (Figure 2). We were fortunate in being able to test this hypothesis using lymphocytes from three patients who had recovered from severe phenytoin hepatotoxicity (16). Cells were

studied after the patients had recovered from the acute episode and a g a i n several months to a y e a r later. Over a phenytoin concentration range of 3 1 - 1 2 5 ~mot/L, assays from each of the patients showed a dosedependent increase in percent of dead cells, while no toxicity occurred in cells from 20 control subjects including patients chronically treated with phenytoin without toxic sequelae. Toxicity to patient cells was dependent on the presence of microsomes and NADPH. Inhibition of epoxide hydrolase in normal cells yielded toxicity curves similar to patient cells in the absence of inhibition of the enzyme. F a m i l y studies revealed interm e d i a t e dose-response curves in cells from parents of the patients and a p a t t e r n in other family m e m b e r s consistent with an autosomal co-dominant p a t t e r n of inheritance (Figure 3). The results indeed suggested t h a t predisposition to phenytoin hepatotoxicity derives from an inherited defect in arene oxide detoxification. Recently, studies on cells from another patient perm i t t e d us to explore further the nature of the detoxification defects (17). The patient had a severe adverse

Cytotoxicity

Hepa t o c e l l u l a r Necrosis

/

Hapten

-Immunologic Reactions -Lymphadenopathy

CLINICAL BIOCHEMISTRY, VOLUME 19, APRIL 1986

143

SPIELBERG reaction to phenytoin including hepatotoxicity and bone marrow aplasia, the latter confirmed by bone marrow biopsy. He recovered, and due to an ongoing seizure disorder, was treated with phenobarbital. The drug was continued for years without toxicity, but because of increasing seizure frequency, his t h e r a p y was switched to carbamazepine. Several weeks into treatment, he again developed hepatotoxicity and bone marrow aplasia. The unusual clinical story raised several questions: 1) Why did he have an adverse reaction to two of the drugs while tolerating the other? 2) Why was bone marrow affected, while t h e other patients had primarily skin and hepatic involvement? Carbamazepine, like phenytoin, is oxidized by cytochrome P-450. One of the metabolites (10,11-epoxide) is a stable compound found in serum and urine of patients treated with the drug, but unstable arene oxide metabolites also are formed. We found that the latter can be cytotoxic in the assay. When we studied cell response to metabolites of the three drugs in the patient's lymphocytes, only metabolites of phenytoin and carbamazepine caused toxicity greater t h a n in control cells. Cells from his mother showed an intermediate response to phenytoin and carbamazepine metabolites, but not to metabolites of phenobarbital. In contrast, cells from several other patients with anticonvulsant-induced hepatotoxicity which we have examined have shown abnormal detoxification capacity for metabolites of all three drugs. The results suggest a heterogeneous group of inherited detoxification disorders, perhaps due to mutations resulting in altered capacity to detoxify different arene oxides. The differences in organ specificity of toxicity m a y relate to different balances of expression of specific cytochromes P-450 and cell detoxification capacity in different tissues, or perhaps differences in immune response to covalently bound drug metabolites.

Discussion Idiosyncratic drug reactions remain a major problem in clinical pharmacology. Genetic heterogeneity within the h u m a n population suggests t h a t polymorphisms will exist for m a n y loci involved in the regulation of drug metabolism which, in turn, will have toxicologic significance. Reactions which can have devastating effects on individual patients and lead to removal of new drugs from the m a r k e t are not easily predicted in animal or early h u m a n trials, and the need for safe approaches to the diagnosis and studying the mechanisms of toxicity is obvious. The in vitro approach we have used for studying adverse reactions to the anticonvulsants has several advantages. Challenge of cells from patients who might have sustained a drug reaction can aid in diagnosis of such reactions. Demonstration of an inherited abnormality in cells from family members helps confirm the diagnosis. If the predictive value of such an approach is verified, counselling of family members about risk of specific types of drug toxicity would be possible. Similarly, prospective in vitro challenge with structural analogs of the offending drug or unrelated compounds with similar efficacy would allow choice of medications 144

with the least risk for peitients who have already had an adverse reaction. Studies of structure: toxicity relationships using the assay may lead to the rational design of drugs for selected patients with pharmacogenetic disorders.

Acknowledgements This work was supported in part by Grant MT-7489 from the Medical Research Council of Canada. S.P.S. is the recipient of a Medical Research Council Scholarship.

References 1. Kramer MS, Leventhal JM, Hutchinson TA, Feinstein AR. An algorithm for operational assessment of adverse drug reactions. JAMA 1979; 242: 623-32. 2. Naranjo CA, Busto V, Sellers EM, et al. A reliable method for estimating the probability of adverse reactions. Clin Pharmacol Ther 1981; 30: 239-45. 3. Weinshilboum, RM, Ed. Human Pharmacogenetics Symposium. Fed Proc 1984; 43: 2295-347. 4. Dhar GJ, Pierach CA, Ahamed PN, Howard RB. Diphenylhydanto;n-induced hepatic necrosis. Postgrad Med 1974; 56: 128-34. 5. Parker WA, Shearer CA. Phenytoin hepatotoxicity: A case report and review. Neurology 1979; 29: 175-8. 6. Siegal S, Berkowitz J. Diphenylhydantoin (Dilantin) hypersensitivity with infectious mononucleosis-like syndrome and jaundice. J Allergy 1961; 32: 447-51. 7. Van Wyck JJ, Hoffmann CR. Periarteritis nodosa: a case of fatal exfoliative dermatitis resulting from '¢Dilantin sodium" sensitization. Arch Intern Med. 1948; 81: 605-11. 8. Robinson HM Jr, Stone JH. Exanthem due to diphenylhydantoin therapy. Arch Dermatol 1970; 191: 462-5. 9. Schreiber MM, McGregor JG. Pseudolymphoma syndrome: a sensitivity to anticonvulsant drugs. Arch Dermatol 1968; 97: 297-300. 10. Eadie MJ, Tyrer JH. Anticonvulsant Therapy: Pharmacological Basis and Practice. Pp. 45-132. 2d ed. New York: Churchill Livingstone, 1980. 11. Jerina DM, Daly JW. Arene oxides: a new aspect of drug metabolism. Science 1974; 185: 573-82. 12. Rao GS, Wortel JP. Gingival metabolism ofphenytoin and covalent binding of its reactive metabolite to gingival proteins. In: Hassel TM, Johnston MC, Dudley KH, Eds. Phenytoin-Induced Teratogenesis and Gingival Pathology. Pp. 189-98. New York: Raven Press, 1980. 13. Martz F, Failinger C III, Blake DA. Phenytoin teratogenesis: Correlation between embryopathic effect and covalent binding of putative arene oxide metabolite in gestational tissue. J Pharmacol Exp Ther 1977; 203: 231-9. 14. Spielberg SP, Gordon GB, Blake DA, Mellits ED, Bross DS. Anticonvulsant toxicity in vitro: Possible role of arene oxides. J Pharmacol Exp Ther 1981; 21Y: 386-9. 15. Woodbury DM, Penry JK, Schmidt RP. Antiepileptic Drugs. New York: Raven Press, 1972. 16. Spielberg SP, Gordon GB, Blake DA, Goldstein DA, Herlong HF. Predisposition to phenytoin hepatotoxicity assessed in vitro. N Engl J Med 1981; 305: 722-7. 17. Gerson WT, Fine DG, Spielberg SP, Sensenbrenner LL. Anticonvulsant-induced aplastic anemia: Increased susceptibility to toxic drug metabolites in vitro. Blood 1983; 61: 889-93. CLINICAL BIOCHEMISTRY, VOLUME 19, APRIL 1986