PHARMACOEPIDEMIOLOGY AND DRUG UTILIZATION

Anticonvulsant usage is associated with an increased risk of procarbazine hypersensitivity reactions in patients with brain tumors

Background: Procarbazine usage in brain tumors has a high incidence of hypersensitivity reactions com- pared with its use in other malignancies. Procarbazine oxidation to a reactive intermediate is enhanced by phenobarbital. Patients with primary brain tumors would have a preferential exposure to anticonvnlsants compared to patients with other malignancies. Objective: To determine whether a&convulsant exposure is associated with procarbazine hypersensitivity reactions in patients with primary brain tumors. Metbouk This retrospective cohort study included 83 patients with primary brain tumors who were treated with procarbazine between 1981 and 1996 at a university hospital-based regional oncology center. Data were extracted by chart review. The data collected included age, sex, race, tumor type, smoking, alcohol usage, and all concomitant medications, as well as creatinine, aspartate aminowferase, total bilirnbin, and anticonvnl- sant sernm levels. Anticonvnlsant exposnre was determined by the presence of detectable serum levels. Cases of procarbazine hypersensitivity reactions were identified through a review of progress notes. Results There were 20 patients with procarbazine hypersensitivity reactions. A significant association between the exposure to anticonvnlsants and the development of procarbazine hypersensitivity reactions was found (p = 0.05). In addition, there was a significant dose-response association between the development of procarbazine hypersensitivity and the presence of therapeutic anticonvnlsant serum levels (p = 0.03). Conclusions: Concomitant exposure to anticonvnlsants is associated with procarbazine hypersensitivity reac- tions, possibly though a reactive intermediate generated by CYP3A isoform induction. All patients in this cohort received enzyme-inducing anticonvulsants. New anticonvnlsants devoid of this property are available. These data support trials that use these newer agents for the prophylaxis of seizures in patients with brain tumors who are to receive procarbazine. (Clin Pharmacol Ther 1997;62:225-9.)

David F. Lehmann, MD, PharmD, Tracy E. Hurteau, BS, Nancy Newman, MS, and Thomas E. Coyk, MD Syracuse, 2V.T.

From the Department of Medicine, the Department of Pharma- cology, the School of Medicine, and the Department of Neu- rosurgery, State University of New York Health Science Cen- ter at Syracuse.

Received for publication Feb. 6, 1997; accepted April 14, 1997. Reprint requests: David F. Lehmann, MD, PharmD, SUNY-HSC

at Syracuse, 750 E. Adams St., UHCC 2123, Syracuse, NY 13210.

Copyright 0 1997 by Mosby-Year Book, Inc 0009-9236/97/$5.00 + 0 13/l/82681

Procarbazine is an alkylating agent with signifi- cant activity in the treatment of brain tumors.‘S2 We previously reported the activity of the procarbazine- containing regimen: methotrexate, vincristine, and procarbazine, in initial adjuvant treatment of glio- mas.3 We found a 34% incidence of hypersensitivity reactions to procarbazine with this regimen, which is consistent with other published observations in gli- oma brain tumors.4’5 This incidence of hypersensi-

225

226 Lehmann et al. CLINICAL PHAFMA COLOGY &THERAPEUTICS

AUGUST 1997

tivity appears to be much higher than that reported in trials of procarbazine-containing regimens for lymphoma, Hodgkin’s disease, and other solid tu- mors, in which such reactions appear to be rare.6

The reasons for the difference in the incidence of procarbazine hypersensitivity in various diseases is unknown. There is no evidence that disease pro- cesses would have any effect on the development of hypersensitivity reactions. Procarbazine is oxidized to a reactive intermediate in vitro.’ Although this reactive intermediate has not been studied in terms of its propensity for production of hypersensitivity, oxidative reactive intermediates from other com- pounds have been implicated in the development of hypersensitivity reactions.* The production of the reactive intermediates of procarbazine and other compounds are enhanced by pretreatment of micro- somes with phenobarbital.7,9

Patients with primary brain tumors that involve the cerebral cortex routinely begin taking anticon- vulsant agents prophylactically upon diagnosis. The most commonly used anticonvulsants (phenytoin, phenobarbital, and carbamazepine) all have well documented enzyme-inducing effects.” Patients with other malignancies in which procarbazine is used would not have the same exposure rate to anticonvulsants. Thus anticonvulsant usage could account for the difference in procarbazine hypersen- sitivity reactions if an intermediate is the culprit compound and if this intermediate is produced in greater amounts in vivo after exposure to enzyme- inducing anticonvulsants. We tested the hypothesis that procarbazine hypersensitivity reactions occur more commonly if used with anticonvulsants in a cohort of patients with primary brain tumors.

METHODS The hospital records of all patients with primary

brain tumors who received procarbazine at the State University of New York Health Science Center (SUNY-HSC) at Syracuse from January 1981 to June 1996 were reviewed. Only those patients who received all cycles of procarbazine at the SUNY- HSC at Syracuse were included in the study.

We recorded the age, sex, ethnic@, and histologic diagnosis of the patients, as well as the available histories of drug allergies, cigarette smoking, and alcohol ingestion at the time of procarbazine admin- istration. All concomitant medications were also re- corded. Serum values of creatinine, total bilirubin, aspartate aminotransferase, and anticonvulsant drug levels were recorded. These chemistries were usu-

ally obtained with the other laboratory tests rou- tinely ordered during the week before administra- tion of chemotherapy. All patients receiving anticonvulsants had serum levels available to record,

Hypersensitivity reactions were identified by a re- view of all progress notes during the dates of pro- carbazine administration to verify the description of the reaction and to confirm procarbazine as the suspected etiologic agent. Hypersensitivity reactions were defined and classified according to the Cancer and Acute Leukemia Group B (CALGB) common toxicity criteria. This classification schema is speci- fied by organ system and graded by severity of re- action from 0 (no toxicity) to 4 (severe toxicity).

In an attempt to investigate the possibility of a dose-dependent relationship between anticonvul- sant exposure and procarbazine hypersensitivity, an- ticonvulsant levels were classified as being either therapeutic or subtherapeutic. The problem of ob- taining multiple anticonvulsant levels for the same patient was addressed in the following manner: For patients with hypersensitivity reactions, the anticon- vulsant level obtained closest to the procarbazine course responsible for the hypersensitivity reaction was classified as therapeutic or subtherapeutic. For nonhypersensitive patients, if an anticonvulsant level was therapeutic at any time, then the patient was included in the therapeutic category. In this manner, any potential bias involving classification of nonhypersensitive patients would be toward the null hypothesis (no dose dependency).

The comparisons of patient characteristics of the two groups were performed with use of the x2 sta- tistic, the Fisher exact test, and the Student t test, as appropriate. Evaluation of differences in outcome between anticonvulsant exposure was assessed with the Mantel-Haenszel x2 statistic. A p value 10.05 was considered to be significant. Analyses were done on a computer with use of CRUNCH (Crunch Corp., Oakland, Calif.) and EPI (USD Inc., Stone Mountain, Ga.) software.

RESULTS Eighty-three patients formed the cohort. All pa-

tients received procarbazine either as a component of a methotrexate, vincristine, and procarbazine combination regimen or as a component of a pro- carbazine, lomustine, and vincristine regimen. The dose of procarbazine was 100 mg/m2/day for 14 days when combined with methotrexate and vincristine. The dose of procarbazine was 60 mg/m2/day for 14 days when combined with lomustine and vincristine.

CLINICAL PHABMA COLOGY &THERAPEUTICS VOLUME 62, NUMBER 2 Lehmann et al. 227

Table I. Patient characteristics*

Variable

Hjpersensitive Nonhypersensihve patients patie& (n = 20) (n = 63)

Age Ot 46.3 ? 14.4 44.3 ? 17.7 Sex (% female) 45% 37% White 100% 97% Glioblastoma (n) 9 36 Astrocytoma (n) 11 22 Oligodenclroglioma (n) 0 4 CNS lymphoma (n) 0 1 Smokers (n = 74) 19% 19% Alcohol users (n = 71) 25% 24% Drug allergies 28% 27% Total bilimbin (mg/dl)t 0.36 5 0.14 0.42 t 0.28 AST (U/L)t 22 L 8.8 20 2 11.3 Creatinine (mg/dl)t 0.94 + 0.2 0.92 k 0.17 Lomustine$ 0 5 Glucocortiakis (n) 19 54 Ketcconazole (n) 1 3 warfarin (n) 2 5 Ciprofloxaein (n) 0 2 Diltiazem (n) 2 2 Fluoxetine (n) 0 1 Phenothiazine (n) 15 43 Propoxyphene napsylate (n) 0 1 Ranitidine (n) 5 7 Verapamil (n) 0 1

CNS, Central nervous system; AST, asp&ate aminotransferase. *No significant differences for any variable listed. ?Values are expressed as mean values + SD. *Administered as a component of procarbazine, lomustine, and vincris-

tine.

Procarbazine was cycled every 4 to 6 weeks accord- ing to the regimen and was administered in 2 to 6 courses, depending on tumor response and toxicity.

Twenty patients received diagnoses of hypersen- sitivity reactions from procarbazine. All 20 patients had a grade 3/4 rash (symptomatic, generalized mac- ulopapular eruption). In addition, one patient had interstitial pneumonitis. In all instances of hypersen- sitivity, procarbazine therapy was discontinued, with subsequent resolution of the findings. No patient who had a hypersensitivity reaction was rechal- lenged with procarbazine. Table I summarizes the demographic and clinical data between the two groups. The groups were comparable for all vari- ables analyzed.

However, as shown in Table II, a significant asso- ciation between the exposure to anticonvulsants and the development of procarbazine hypersensitivity was found (p = 0.05). Exposure was documented by the presence of detectable anticonvulsant serum lev- els. The anticonvulsants used were phenytoin, phe-

Table II. Anticonvulsant exposures and procarbazine hypersensitivity

Hypersensitive Nonhypersensitive patients (n) patients (n)

Anticonvulsant exposure*

No anticonvulsant exposure

Therapeutic anticonvulsant levelt$$

Subtherapeutic level or unexposed

19 47

1 16

16 33

4 30

*D = 0.05. fp = 0.03. $Phenytoin (10 to 20 &ml), phenobarbital (15 to 40 pgiml), carba-

mazepine (8 to 12 &ml), and valproic acid (SO to 100 kg/ml). SNo toxic levels found.

nobarbital, carbamazepine, and valproic acid. Val- proic acid (a non-enzyme inducer) was used only in combination therapy. Subgroup analyses that com- pared specific anticonvulsant combinations and mul- tidrug regimens versus monotherapy did not show significant differences.

The results after the classification of anticonvul- sant serum levels into therapeutic and subtherapeu- tic groups in the exposed patients are shown in Table II. A significant association between the pres- ence of therapeutic serum levels of anticonvulsants and the occurrence of procarbazine hypersensitivity reactions was found @ = 0.03).

DISCUSSION We demonstrated a significant association be-

tween anticonvulsant exposure and the occurrence of hypersensitivity reactions to procarbazine in a cohort of patients treated for brain tumors at our institution. The association is strengthened by the demonstration of potential dose dependency as an- alyzed by the presence of therapeutic anticonvulsant levels in relationship to the occurrence of procarba- zine hypersensitivity. This method was chosen be- cause it provides a convenient way to compare dos- ing intensity between various agents with inherent differences in pharmacokinetics and pharmacody- namics. Furthermore, it circumvents problems asso- ciated with questions of compliance.

The patients were comparable in demographics and tumor type. Other variables that have been shown to effect drug metabolism, including ethnic- ity,” ingestion of alcohol,‘* cigarette smoking,13 and

228 Lehmann et al.

concomitant microsomal enzyme inducers or inhib- itors,i4 were also comparable in the cohort.

Enzyme-inducing anticonvulsants have been shown to affect both the pharmacokinetics and phar- macodynamics of other chemotherapeutic agents in both preclinical and clinical studies. Experiments that used a mouse model intracerebrally implanted with the 9L gliosarcoma cell line showed that long- term oral pretreatment with phenobarbital acceler- ated the plasma clearance of carmustine, decreasing both its antitumor effects and its systemic toxicity.” Mice pretreated with either phenytoin or phenobar- bital showed protection from the myelosuppressive effects of busulfan treatment.r6

Clinical studies have shown increased plasma clearance of teniposide and etoposide in patients concomitantly treated with enzyme-inducing anti- convulsants.17’18 Two preliminary reports concern- ing the interactions of anticonvulsants with the phar- macokinetics of paclitaxel have recently been published. 19,20 The first report noted that the maxi- mally tolerated dose and plasma clearance of pacli- taxel were markedly increased in patients with re- current brain tumors who received anticonvulsants compared with those not who did not receive anti- convulsants.19 The second showed that steady-state concentrations and toxicity of paclitaxel were signif- icantly decreased in patients who received 96hour infusions of paclitaxel for newly diagnosed gliomas by concomitant exposure to enzyme-inducing anti- convulsants.20

Procarbazine is a pro-drug that is activated to cytotoxic compounds through unspecified cyto- chrome P450 isoform(s).21 Our results indicate that procarbazine is also metabolized to species respon- sible for hypersensitivity phenomena by a cyto- chrome P450 isoform(s) that is inducible by certain anticonvulsants. Because phenobarbital, phenytoin, and carbamazepine all induce CYP3A, it is possible that these isoforms produce the metabolite respon- sible for procarbazine hypersensitivity reactions.22 Our findings may prompt investigations into whether manipulation of this subfamily alters the hypersensitivity response to procarbazine, as shown for other compounds that are known to generate reactive intermediates.23

In summary, our findings add potential insight into the cause for the greater rate of procarbazine hypersensitivity reactions observed in patients with brain tumors as opposed to other malignancies for which procarbazine is used. All patients who were exposed to anticonvulsant agents in this cohort were

CLINICAL PHARMACOLOGY &THERAPEUTICS AUGUST 1997

given enzyme inducers. However, anticonvulsants are available that do not have this property.24 Our results provide the basis for trials that use non- enzyme-inducing anticonvulsants for the prophy- laxis of seizures in brain tumor patients when pro- carbazine therapy is anticipated.

References 1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

Rodriguez L, Prados M, Silver P, Levin VA. Reeval- uation of procarbazine for the treatment of recurrent malignant central nervous system tumors. Cancer 1989;64:2420-3. Newton HB, Junck L, Bromberg J, Page MA, Green- berg HS. Procarbazine chemotherapy in the treat- ment of recurrent malignant astrocytomas after radi- ation and nitrosourea failure. Neurology 1990;40: 1743-6. Coyle T, Baptista J, Winfield J, Clark K, Poiesz B, Kirshner J, et al. Mechlorethamine, vincristine and procarbazine chemotherapy for recurrent high-grade glioma in adults: a phase II study. J Clin Oncol 1990; 8:2014-g. Coyle T, Bushunow P, Winfield J, Wright J, Graziano S. Hypersensitivity reactions to procarbazine with MOP chemotherapy for glioma. Cancer 1992;69:2532- 40. Cairncross G, MacDonald D, Ludwin S, Lee D, Cas- cino T, Buckner J, et al. Chemotherapy for anaplastic oligodendroglioma. J Clin Oncol 1994;12:2013-21. Longo D, Young R, Wesley M, Hubbard SM, Duffey PL, Jaffe ES, et al. Twenty years of MOPP therapy for Hodgkin’s disease. J Clin Oncol 1986;4:1295-306. Sinha BK. Cytochrome P-450 and peroxidase depen- dent activation of procarbazine and iproniazid in mammalian cells. Free Radic Res Commun 1991;lS: 189-95. Park BK, Kitteringham NR. Drug protein conjugation and its immunological consequences. Drug Metab Rev 1990;22:87-144. Cribb AE, Spielberg SP. Hepatic microsomal metab- olism of sulfamethoxazole to the hydroxylamine. Drug Metab Dispos 1990;18:784-7. Smith MC, Bleck TP. Convulsive disorders: toxicity of anticonvulsants. Clin Neuropharmacol 1991;14:97- 115. Relling MV, Lin JS, Ayers GD, Evans WE. Racial and gender differences in N-acetyltransferase, xan- thine oxidase, and CYPlA2 activities. Clin Pharmacol Ther 1992;52:643-58. Hobbs WR, Rail TW, Verdoorn TA. Hypnotics and sedatives: ethanol. In: Hardman JG, Limbird LE, Mo- linoff PB, Ruddon RW, Goodman Gilman A, eds. Goodman and Gilman’s the pharmacological basis of therapeutics. New York: McGraw-Hill; 1996:361-96. Kalow W, Tang BK. Caffeine as a metabolic probe:

CLINICAL PHARhU COLOGY & THERAPEUTICS VOLUME 62, NUMBER 2 Lehmann et al. 229

exploration of the enzyme-inducing effects of ciga- rette smoking. Clin Pharmacol Ther 1991;49:44-8.

14. Slaughter RL, Edwards DJ. Recent advances: the cytochrome P450 enzymes. Ann Pharmacother 1995; 29:619-24.

15. Levin VA, Steams J, Byrd A, Finn A, Weinkam RJ. The effect of phenobarbital pretreatment on the an- titumor activity of 1,3-bis(2-chloroethyl)- 1-nitrosourea (BCNU), 1(2-chloroethyl)-3-(2,6-dioxo-3-piperidyl-l- nitrosourea (PCNU), and on the plasma pharmacoki- netics and biotransformation of BCNU. J Pharmacol Exp Ther 1979;208:1-6.

16. Fitzsimmons WE, Ghalie R, Kaizer H. The effect of hepatic enzyme inducers on busulfan neurotoxicity and myelotoxicity. Cancer Chemother Pharmacol 1990;27:226-8.

17. Baker DK, Relling MV, Pui CH, Christensen ML, Evans WE, Rodman JH. Increased teniposide clear- ance with concomitant anticonvulsant therapy. J Clin Oncol 1992;10:311-5.

18. Mross K, Bewermeier P, Kruger W, Stockschlader M, Zander A, Hossfeld DK. Pharmacokinetics of undi- luted or diluted high dose etoposide with or without busulfan administered to patients with hematologic malignancies. J Clin Oncol 1994;12:1468-74.

19. Fete11 MR, Grossman SA, Fisher J, Erlanger B, Stockel J, Piantodosi S. Preirradiation paclitaxel in glioblastoma multiforme (GBM): efficacy, pharmacol- ogy and drug interactions [abstract]. Proc Am Sot Clin Oncol 1996;15:153a.

20. Chang S, Schold C, Spence A, Berger M, McAllister L, Mehta M, et al. Phase I/II study of paclitaxel chemotherapy for recurrent malignant glioma: pre- liminary report of North American Brain Tumor Con- sortium [abstract]. Proc Am Sot Clin Oncol 1996;15: 153a.

21. Tweedie DJ, Erikson IM, Prough RA. Metabolism of hydrazine anticancer agents. Pharmacol Ther 1987;34: 111-27.

22. Shen WW. Cytochrome P450 monooxygenases and interactions of psychotropic drugs: a five year update. Int J Psychiatry Med 1995;25:277-90.

23. Cribb AE, Spielberg SP, Griffin GP. N4- Hydroxylation of sulfamethoxazole by cytochrome P450 of the cytochrome P450 2C subfamily and re- duction of sulfamethoxazole hydroxylamine in human and rat hepatic cytochromes. Drug Metab Dispos 1995;23:406-14.

24. Dichter MA, Brodie MJ. New antiepileptic drugs. N Engl J Med 1996;334:1583-90.