TOXICOLOGY

AND

APPLIED

PHARMACOLOGY

Phosphoramide

DAVID Department

oJObstetrics Little Rock.

107,

472-48

1 ( 199 1)

Mustard Is Responsible for the Ovarian Toxicity of Cyclophosphamide R. PLOWCHALK’

AND DONALD

R. MATTISON’

and Gynecology and Division of‘ Toxicology, Universit?, oj‘ Arkansas Arkansas 72205; and Division of Developmental and Reproductive Nutional Center,for To.yicological Reseurch. Jefferson, .4rkansas

Received

June 14. 1990: accepted

October

for

Medical

Sciences.

Toxicology

22. 1990

Phosphor-amide Mustard Is Responsible for the Ovarian Toxicity of Cyclophosphamide. PLOWD. R., AND MATTISON. D. R. ( I99 1). Toxicol. Appl. Pharmacol. 107,472-48 1. Although cyclophosphamide (CPA) is an ovarian toxicant, the responsible metabolite(s) have not been identified. The purpose of these experiments was to determine if phosphoramide mustard or acrolein were the proximate toxicants produced by metabolic activation of CPA. To do this analogs of CPA known to generate either phosphoramide mustard or acrolein in vivo were assessed for their ability to produce ovarian toxicity as measured by differential follicle destruction, ovarian volume loss, and uterine weight loss and compared to the effectsproduced by CPA. Phosphoramide mustard cyclohexylamine salt (PMC) and trans-4-phenylcyclophosphamide (T4P). both of which generate phosphoramide mustard. and didechlorocyclophosphamide (DCPA) and ally1 alcohol (AA) which generate acrolein were administered ip to female C57BL/6N mice, IO- 12 weeks old, at doses equimolar to 0, 25, 75, 200. or 500 mg/kg of CPA. Three days later the animals were killed, their uterine weights measured and their ovaries removed, fixed, and serially sectioned. Only PMC and T4P produced ovarian toxicity. On an equimolar basis these compounds were over twice as potent as CPA. Both caused a significant reduction in uterine weight (to 50% of controls) at doses of 200 (PMC) and I50 mg/kg (T4P). PMC and T4P also caused a 50% reduction in ovarian volume at doses above 75 mg/kg. Primordial follicles were most sensitive: EDSOs were 76.9, 25.3, and 19.3 mg/kg (0.276, 0.091. and 0.069 mmol/kg) for CPA. PMC. and T4P. respectively. Growing follicle numbers were also reduced by T4P and PMC. an effect not seen with CPA treatment. Finally, antral follicles were significantly reduced by all doses of PMC. and with T4P at doses greater than 75 mg/kg. The highest doses of PMC. T4P. and CPA all caused a reduction in antral follicle numbers to less than one percent of controls. Didechlorocyclophosphamide (DCPA) and ally] alcohol (AA), compounds that generate acrolein but not phosphommide mustard in vivo, had no effect on any of the parameters measured even when injected directly into the ovary. This suggeststhat phosphoramide mustard is responsible for CPA ovarian toxicity. The greater potency of PMC and T4P compared to CPA is likely the result of these compounds bypassing important detoxification steps, therefore, more of the parent compound reaches the ovary as the toxic metabolite. 0 1991 Academic Rest, IN. CHALK,

Various degrees of ovarian dysfunction are commonplace among women treated with the antineoplastic agent cyclophosphamide

(CPA). Warne et al. (1973) and Koyama et al. (1977) found a high percentage (>75%) of women receiving CPA therapy developed amenorrhea that was often irreversible and led to premature ovarian failure. It has previously been suggested that this toxicity is the result of chemical injury to specific ovarian components. In particular, results from rodent studies indicate that primordial and antral

’ Current address: RJ Reynolds Tobacco Co., Toxicology Research Division, Building 630-2, Winston-Salem, North Carolina 27 102 2 To whom correspondence and reprint requests should be addressed at Room 1 I 1, Parran Hall, Graduate School of Public Health, 130 Desoto Street, Pittsburgh, PA 1526 I. 0041-008X/91

$3.00

Copyright 0 1991 by Academic Press. Inc. All rights of reproduction in any form reserved.

472

CYCLOPHOSPHAMIDE

follicles are extremely sensitive to CPA and are destroyed in a time- and dose-dependent fashion (Mattison et al., 1983a,b; Plowchalk and Mattison, 1988; Miller and Cole, 1970; Shiromizu et al., 1984). Follicle destruction is also accompanied by depressed estradiol levels, increases in follicular atresia, degenerative morphometric changes in the ovary (follicles and corpora luteum) and uterine weight loss (Jarrell et al., 1987; Plowchalk et al., 1989). CPA itself is essentially devoid of any biological activity without metabolic activation (Brock, 1967). Upon biotransformation, CPA yields a number of metabolites with various cytotoxic activity. As illustrated in Fig. I, the first step of CPA metabolism is hydroxylation by cytochrome P450 to form 4-hydroxycyclophosphamide, which can be either detoxified to the nonreactive metabolite 4-ketocyclophosphamide (Brock, 1976) or undergo ringopening to form aldophosphamide. Aldophosphamide has several fates. It can be detoxified by cytosolic enzymes to either the principal urinary metabolite carboxycyclophosphamide or the minor metabolite alcophosphamide. Alternatively, aldophosphamide can spontaneously release the therapeutically active metabolite phosphoramide mustard (PM) (Colvin et al., 1976) and the highly reactive c+unsaturated aldehyde acrolein (ACR). Both of these metabolites are potent cytotoxic agents which exert their biological effects through covalent binding to important cellular macromolecules. The molecular binding site of PM is principally DNA (Gurtoo et al., 1978) where it forms N-7 guanosine adducts by alkylation (Mehta et al., 1980). Because PM is a bifunctional agent, it can cause both DNA lesions associated with binding at a single site on the DNA strand (depurination, abnormal basepairing, strand scission) and also cause lesions associated with alkylation at two sites (DNADNA, DNA-protein cross-links). Both of these are thought to be important mechanisms of PM cytotoxicity (Mirkes et al., 1985) and are responsible for its mutagenic (Winckler et al.,

OVARIAN

473

TOXICITY

1984), clastogenic (Wilmer et al., 1986), teratogenic (Mirkes, 1985), and antiproliferative properties. The toxic effects attributed to ACR are reported to be the result of covalent binding to critical protein sulthydryls (Beauchamp et al., 1985). This may result in damage to important enzyme systems as demonstrated by Gurtoo et al., (1981a) and Marine110 et al., (1981) who found that ACR can denature cytochrome P450 and inhibit P450 reductase. CPA-derived ACR has also been shown to interact with membrane proteins (Wildnauer and Oehlmann, 1982) reduce hepatic thiol concentrations (Gurtoo et al., 198 lb; Plowchalk and Mattison, 1989b) and cause hemorrhagic cystitis (Plotz et al., 1979). Considering the chemical reactivity of PM and ACR, we have investigated the relative roles of these metabolites in CPA-induced ovarian toxicity. To do this, structural analogs of CPA or CPA metabolites that selectively generate either PM or ACR were assessed for their ability to induce toxic changes in the mouse ovary normally produced by CPA treatment. These endpoints included differential follicle destruction, changes in ovarian volume, and uterine weight loss. The four compounds investigated in these studies are illustrated in Fig. 2 and include didechlorocyclophosphamide (DCPA), ally1 alcohol (AA), which will produce ACR upon metabolism, and trans-4-phenylcyclophosphamide (T4P) and phosphoramide mustard cyclohexylamine salt (PMC), which will generate PM. MATERIALS

AND METHODS

Chemicds. Cyclophosphamide and ally1 alcohol were purchased from Sigma Chemical Co. (St. Louis, MO). Acrolein was purchased from Aldrich Chemical Co. (Milwaukee, WI). Phosphoramide mustard (cyclohexylamine salt) (NSC-69945) was donated by the National Cancer Institute, Drug Synthesis & Chemistry Branch, Division of Cancer Treatment (Bethesda, MD). Didechlorocyclophosphamide was a generous Sift from Dr. Ulf Niemeyer of Asta Pharma, Bielefeld, West Germany and truns-C phenylcyclophosphamide was synthesized and kindly donated by Dr. Susan Ludeman, Catholic University (Washington, DC). All compounds were used as obtained from

PLOWCHALK

474

AND MATTISON

O N--CH*, CICH,CH “~-6’~ C,, CICH CH’ ‘O-CH’ * CY&[;PHOSPHAMlD?E

0 ’ N-E, N-k’H CH. ClCH$H/ ‘O-CH( z 4-KETOCYCLOPHOSPHAMIDE CICH,CH,,

1

II PIGuYnE

OH

CICH,CH,’ ‘O-a-l/ &HYDROXYCYCLOPHOSPHAMlDE CtCH,CH4

II

ALDEHYOE

IN% ,N-P,=O

o z II

I

DEHYDROGENASE

CICH,CH 4

/ NH2 N-P=0 CICH,CH/ ‘OCH CH !-OH CARSOXYCYCLOPHO:PH~MtDE

CICH,CH,+

/NH, 0 N-P=0 CH+H-;H ClCHp$’ ‘OH PHOSPHORAMIDE MUSTARD ACROLEIN FIG. 1. Pathway for the bioactivation and detoxification of cyclophosphamide

their respective supplier. Purity was assured by thin layer chromatography which demonstrated a single band for all chemicals used in these experiments. Animak Female C57BL/6N mice IO-12 weeks of age were obtained from the National Center for Toxicological Research breeding colonies. The mice were housed 3 or 4 per cage, allowed food and water ad libitum. and kept on a I2-hr/ 12-hr, light/dark cycle. Dose-response studies. Each compound was dissolved in a 1:I mixture of saline:DMSO and used within I hr of

preparation. To determine the dose-response of follicle destruction and ovarian morphometric changes, mice were administered a single intraperitoneal injection (0.1 ml) of a given analog at a dose equimolar to either 25, 75, 200, or 500 mg/kg of cyclophosphamide. Each dose group contained five animals which were compared to a single vehicle control group (n = 5) that received a 0.1 ml injection of the saline:DMSO vehicle. Three days following treatment the mice were killed by cervical dislocation at which time body weight and uterine weight were measured and ovaries

YH

2 CH 3CH /

\O-CH,/

CH2=CH-CH,

Dldechlorocvclophosphan,ide

Ally1

[DCPA)

CICH,CH ?\

/ NH2

CICH,CH *\

N-P=0 CICH,CH / Phosphoramide

’ OH Mustard

(PM)

CICH,CH

Alcohol

(AA)

00

* N-k~, N - VH

/

Trans-.L-phenylcyrlophosphamlde

CH 2

‘o-CH,/ (T1P)

FIG. 2. Chemical structures of compounds used to determine the active metabolite(s) of CPA responsible for ovarian toxicity.

CYCLOPHOSPHAMIDE were collected for sectioning. This time point was chosen because we have previously determined that the effects of CPA on the reproductive system in this animal model are maximal at 3 days (Plowchalk and Mattison, 1989a). Intraovarian Injection (IOI) Experiment. To avoid possible dispositional differences between the compounds, direct intraovarian injections (101) of each analog was used as a second treatment method, with uterine weight changes as the endpoint of toxicity. This procedure (101) has been documented by Takizawa ef al. (1984) as an effective method for directly exposing the ovary to a chemical and for evaluating intra- and extraovarian activation of xenobiotics. Under light ether anaesthesia, a dorsolumbar paraspinal incision was made through the skin. Pointed forceps were used to puncture the retroperitoneal musculature, making a hole through which the ovary and uterine horn were exteriorized. After the uterus was lightly clamped just below the ovary, I ~1 of the respective compound (0.5 pmol/pl) was injected into the ovary using a IO-PI Hamilton syringe (Hamilton Co., Reno, NV). The uterus was then unclamped and the ovary replaced in the peritoneal cavity. This procedure was performed on both ovaries. The incisions were closed with wound clips and the animals were allowed to recover in cages warmed to 38°C. The compounds used for the 101 experiment were prepared as follows. Cyclophosphamide, phosphoramide mustard, ally1 alcohol, and ACR were dissolved in saline to a concentration of 0.5 rmol/rl. Because didechlorocyclophosphamide and trans-4-phenylcyclophosphamide are relatively water insoluble, they were prepared in a 1: 1 mixture of DMSO:saline to the same concentration as above. All comparisons were made against the appropriate vehicle controls. The dose injected into each ovary (0.5 pmol/rl) was estimated from results ofa previous [‘H]CPA tracer study ([chloroerhyl-3H]cyclophosphamide, purity 98+% used as obtained from Amersham Corp.. Arlington Heights, IL, unpublished data, Plowchalk and Mattison). Briefly. the amount of radiolabeled CPA in the ovary (DPM/mg of ovary) after a 500 mg/kg ip injection of CPA was determined from I - 168 hr. A concentration time curve was used to estimate the amount of CPA in the ovary at time zero (C,). This concentration (DPM/mg) was then multiplied by the ovarian weight (I’,) to obtain the dose for 101. Histological preparation. Immediately following the animal’s death, ovaries were removed, placed in Bouin’s fixative, and then transferred to 70% ethanol after 24 hr. The tissue was then dehydrated in ethanol, embedded, serially sectioned (6.0 pm), and stained with hemotoxylin and eosin. D&rential follicle counts. Differential follicle classifications were made using a modification of the method by Pedersen and Peters (I 968) where types I -3b, types 45b, and types 6-8 were grouped as primordial, growing, and antral follicles, respectively. Only follicles that had oocyte chromatin visible in the cross section were counted,

OVARIAN

475

TOXICITY

and no attempt was made to differentiate atretic from normal follicles. Total follicle numbers were calculated by summing the follicle counts obtained from every 10th section. The numbers reported are not absolute ovarian follicle numbers but are relative to the counting procedure. Doseresponse curves and EDSOs for primordial follicle destruction were estimated with a four-parameter logistic function Y=----+d

a-d

I + (X/C)b

’

where Y is the follicle number. X is the dose, a is the follicle number when X = 0, dis the follicle number when X is infinite, c is the ED50, and b is the slope (De Lean ef al., 1978) using NLIN nonlinear curve fitting software (V. Singh, Cranford, NJ). Morphometric analysis. Ovarian volume and component volumes were determined with the Bioquant IV Morphometric System (R & M Biometrics, Inc., Nashville, TN) which consisted of a Leitz Orthoplan Z microscope (Leitz Inc., Rockleigh, NJ), an MTI-65 monochromatic video camera (Dage-MTI, Michigan City, IN), a Summagraphics digitizing tablet (Summagraphics, Fairfield, CT), and a Dell 286 microcomputer (Dell Computer Corp., Austin, TX). The area of every tenth ovarian section was measured and then multiplied by the section thickness (6.0 pm) to obtain the section volume. Each section volume was then multiplied by 10 (number of sections skipped) and all of these volumes were summed and taken to represent total ovarian volume. Statistics. All data was analyzed with VMS SAS Version 5.16 statistical software (SAS Institute Inc., Gary, NC) on a VAX 8650 computer. Analysis of variance between doses were made with a standard ANOVA procedure or Genera1 Linear Model (GLM) Procedure (SAS Users Guide, 1982) and analysis of the means was performed with either Duncan’s multiple range test or t tests.

RESULTS Dose-Response of Follicle Destruction CPA caused a significant reduction in both primordial and antral follicles, but had no effect on the growing follicle pool (Table 1). Control primordial follicle numbers (3 10 f 45) were reduced by 1549, 87, and 99 percent following treatment with CPA at doses of 2575,200, and 500 mg/kg, respectively. The ED50, i.e., the dose required to destroy 50% of the follicles, was estimated to be 76.9 mg/ kg (0.276 mmol/kg) when fit to a sigmoidal dose-response model (Fig. 3). The effects of CPA on the antral follicle pool were more

PLOWCHALK

476

AND MATTISON

TABLE I DIFFERENTIAL FOLLICLE COUNTS AND BODY WEIGHTS 3 DAYS AFTER TREATMENT WITH A SINGLE INTRAPERITONEAL DOSE OF CYCLOPHOSPHAMIDEOR ANALOGS Follicle number’ Treatment

Dosea

Body weight’

Antral

Vehicled CPA

0 25 75 200 500 25 75 200 500 25 75 200 500 25 75 200 500 25 75 150

19.6 2 0.6 19.1 t 0.6 18.8 + 0.3 17.5 k 0.6’ 17.3 * 0.3’ 18.9 k 0.9 19.1 + 0.4 19.0 -t 0.5 18.4 t 1.1 19.8 zk 0.6 20.6 + 0.5 20.3 -+ 0.7 20.9 f 0.5 18.3 + 0.3 16.3 ? 0.3’ 14.5 + 0.6’ 13.9 + 0.3’ 18.5 f 0.7 18.3 -+ 0.4 17.4 + 0.5’

8.6 f 0.4 6.0 f 1.1 4.0 * 1.1’ 7.0 * 1.0 0.6 f 0.7’ 6.2 f 1.2 8.0 + 1.4 5.4 * 1.5 7.6 + 1.5 9.6 f 1.6 5.6 + 1.5 8.0 + 0.4 7.2 f 0.9 4.8 + 0.6’ 3.8 + 0.9’ 0.5 * 0.3’ 0.8 f 0.8’ 10.4 2 2.0 5.2 f 1.4’ 0.2 t 0.2’

DCPA

AA

PMC

T4P

Growing 48.4 i 48.0 ?I 34.0 f 45.6 + 41.4 iz 34.0 iz 56.0 f 41.4 Ii 58.2 -I 53.4 f 33.0 f 47.4 f 51.6 + 54.7 f 34.2 + 23.0 + 25.6 t 58.4 f 42.0 f 23.8 f

9.7 12.0 4.5 8.5 5.3 3.1 3.9 6.8 4.7 4.6 4.9 7.0 5.4 6.7 5.4 4.1’ 6.9’ 6.6 8.8 4.7’

Primordial 3 10.8 k 45.5 263.8 +- 34.9 159.3 k 23.3’ 42.4 k 7.1e 1.0 rt 0.5’ 255.8 f 24.2 3 16.2 + 29.9 258.2 F 58.4 348.8 + 44.2 246.4 f 57.1 219.6 -t 29.1 246.0 + 18.9 254.8 k 26.6 156.3 k 29.6’ 7.2 a 0.9’ 2.3 t 0.8’ 1.2 * 0.8’ 7 I .2 + 32.4’ 0.0 * 0.0’ 0.6 + 0.6’

Note. Follicle numbers were determined in one ovary that was serially sectioned (6 pm thick). Each follicle type (primordial, growing, and antral) was counted in every tenth section and counts from all sections were summed for total follicle number. a Doses equimolar to these doses of cyclophosphamide (mg/kg). ’ Mean body weight (grams) + SEM, (n = 5). ’ Mean follicle number + SEM, (n = 5). d DMSO/saline. 1:I ’ Significantly different from controls at p i 0.05, ANOVA, and Duncan’s multiple range test.

variable, with the most pronounced destruction of follicles (7% of controls) at the highest dose. PMC produced a dose-dependent destruction of all follicle types. Primordial follicles were most sensitive (ED50 = 25.3 mg/kg, 0.091 mmol/kg) and were reduced by 98% with a dose as low as 75 mg/kg. Unlike CPA, PMC was toxic to the growing follicle pool and destroyed 47% and 53% percent of the follicles with doses of 200 and 500 mg/kg, respectively. Antral follicles were also significantly reduced by PMC at all doses. The analog T4P produced overt systemic toxicity at doses of 200 and 500 mg/kg, there-

fore the highest dose used was 150 mg/kg. As seen with CPA and PMC, T4P exhibited the greatest toxicity to the primordial follicles in a dose-dependent manner; however, the doseresponse curve and the ED50 calculated for this compound are unreliable since only one point fell on the curve (Fig. 3). The growing follicles were also reduced at the highest dose (150 mgfkg) to 49% of controls. The 200 and 500 mg/kg doses also lead to a loss in antral follicles to 60 and 2% of controls, respectively. DCPA and AA had no effect on primordial, antral, or growing follicles at any doses tested (Table 1 and Fig. 4).

CYCLOPHOSPHAMIDE

l CPA (ED59 = 76.9 mg/kd 0 PMC (ED50 = 25.3 ma/kg) . T4P (ED50 = 19.3 m&q)

O----m 10

OVARIAN

I 1

Y

-10

1000 Dose

TOXICITY

Dose

(q/kg)

100 (rr,g/kg)

1000

FIG. 3. Dose-response curves for primordial follicle destruction by cyclophosphamide (CPA), phosphoramide mustard cyclohexlyamine salt (PMC), and truns-4-phenylcyclophosphamide (T4P). Each point on the curve represents the mean follicle number of five mice + SEM. The curves were fit to the four parameter logistic function as described by De Lean et al. (1978) using NLIN (V. Singh, Cranford, NJ). The ED5Os estimated with this dose-response model were 76.9, 25.3, and 19.3 mg/kg (0.276, 0.09 I, and 0.069 mmol/kg) for CPA, PMC, and T4P, respectively.

FIG. 4. Dose-response curves for primordial follicle destruction by cyclophosphamide (CPA), didechlorocyclophosphamide (DCPA), and ally1 alcohol (AA). DCPA and AA were not significantly different from controls at any dose tested. No attempt was made to fit the AA or DCPA data to the sigmoidal dose-response model described in Fig. 3.

Ovarian Volume

Three days following an 101 of CPA or the other test compounds, only PMC and T4P were found to have caused a significant reduction in both ovarian and uterine weight, whereas CPA, DCPA, AA, and ACR had no effect (Fig. 6).

The effect of CPA and the other test compounds on ovarian volume are shown in Fig. 5. Only PMC and T4P caused an ovarian volume loss similar to that seen with CPA treatment, whereas DCPA and AA had no effect. PMC and T4P were also more potent than CPA in producing this effect. PMC reduced the ovarian volume by 26,6 1, and 6 1 percent at 75, 200, and 500 mg/kg, respectively, and 150 mg/kg of T4P reduced the volume by 52%. The greatest volume loss observed with CPA was 50% (500 mg/kg). Uterine Weight The uterine weight loss produced by each compound studied is listed in Table 2. Control uterine weight (87.9 f 17.6 mg/20 g body weight) was significantly reduced by CPA (500 mg/kg) by 54%, PMC (500 and 200 mg/kg) by 50%, and T4P (150 mg/kg) by 48%. DCPA and AA had no effect on uterine weight at any of the doses tested.

Intraovarian

,i%;

Injections (101) Experiment

1

siz2.5 2.0

5

1.5

% b

1.0 0.5 0.0

FIG. 5. Dose-dependent changes in ovarian volume after treatment with a single ip injection of DMSO/saline (I:]) (VEH), cyclophosphamide (CPA), phosphoramide mustard cyclohexylamine salt (PMC), trans-4-phenylcyclophosphamide (T4P), ally1 alcohol (AA), and didechlorocyclophosphamide @CPA). Each compound was given at doses equimolar to 25, 75. 200, or 500 mg/kg CPA. Each bar is the mean ovarian volume (mm)) + SEM of five mice. *Statistically different from VEH at p < 0.05 as determined by ANOVA and t test.

478

PLOWCHALK

AND MATTISON

TABLE 2 DOSE-DEPENDENT~HANGESIN MOUSEUTERINEWEIGHT~DAYSFOLLOWINGINTRAPERITONEAL OFCYCLOPHOSPHAMIDE,DIDECHLOROCYCLOPHOSPHAMIDE,ALLYLALCOHOL,PHOSPHORAMIDE HEXYLAMINE SALTANDITQ~S-4PHENYLCYCLOPHOSPHAMtDE

Compound

Dose (mg/kg)“:

Cyclophosphamide Didechlorocyclophosphamide Ally1 alcohol Phosphoramide mustard cyclohexylamine salt [ truns-4-Phenyll-cyclophosphamide

25

87.8 t 17.2h (99.9)c 96.9 f 23.6 (I 10.2) 82.5 f 12.9 (93.9) 91.3 + 19.9 (103.4) 89.9 f 10.9 (102.3)

ADMINISTRATION MUSTARD~YCLO-

15

200

500

66.5 * 8.3 (75.7) 79.8 * 11.4 (90.8) 12.4 -+ 14.0 (82.4) 56.5 + 8.4 (64.3) 76.7 + 8.5 (87.3)

92.1 IL 9.9 (104.8) 100.9 ?I 21.3 (114.8) 81.9 + 14.2 (93.2) 44.6 f 2.2d (50.7) 45.9 f 3.3d,’ (52.2)

40.6 + 2.3d (46.2) 79.8 * 19.0 (90.8) 84.0? 11.5 (95.6) 43.8 f 3.1d (49.8)

a Doses equimolar to these doses of cyclophosphamide. b Uterine weight (mg/20 g body weight) + SEM. (n = 5 or 6). ’ Change in uterine weight expressed as a percentage of vehicle control uterine weight (87.9 +_ 17.6 mg/20 g body weight). d Significant at p < 0.05, ANOVA, and Duncan’s multiple range test. ’ Dose was I50 mg/kg.

DISCUSSION In this study, treatment with CPA resulted in significant destruction of both primordial and antral follicles, ovarian atrophy, and uterine weight loss. These biomarkers have been previously shown to reflect ovarian damage induced by xenobiotics (Mattison et al., 1983, 1989; Plowchalk and Mattison, 1989a). They are also indicative of specific changes in ovarian function and predictive of future reproductive competence. Primordial follicle counts reflect the degree to which the reproductive lifespan is reduced because this follicle pool is finite and the source of all other follicles (Mattison et al., 1987). Reductions in growing and antral follicle numbers indicate a xenobiotic can disrupt folliculogenesis and produce infertility. Ovarian volume measurements reflect not only alterations in the larger follicle types, but also changes in corpora lutea and ovarian interstitial tissues. Finally, uterine weight changes are an indirect measure of ovarian steroid production, in particular estradiol from the antral follicles.

The compounds T4P and PMC, which both generate PM but not ACR, were used to determine if ovarian toxicity was caused by CPAderived PM. The direct use of PM enabled us to observe the effects of this metabolite on the ovary, while eliminating the need for bioactivation and excluded any toxicity that might be associated with ACR. However, several investigators have suggested that extracellularly generated PM is poorly permeable to cells since it is charged at physiologic pH (Boyd et al., 1986). They suggest that the metabolites 4-hydroxycyclophosphamide and aldophosphamide (AP) are important transport forms of CPA which enter the target cell, fragment, and release PM and ACR. Therefore, in addition to PMC, T4P was used because it forms the acyclic metabolite phenylketophosphamide (PKP) which can enter the cell as does AP and fragment to PM. This pathway closely resembles that seen with CPA. In fact the decomposition half-lives of AP and PKP are similar (66 and 77 min., respectively; Ludeman et al., 1986). Both PMC and T4P were found to produce extensive ovarian damage, reducing ovarian

CYCLOPHOSPHAMIDE

2 8 2

125 100

& &

75

2 .c? ; 2 F o

50 25 0

CPA

PMC

T4P

DCP

ACR

AA

6. Ovarian and uterine weight of C57BL/6N mice 3 days after intraovarian injections (0.5 rmol/ovary) of cyclophosphamide (CPA), phosphoramide mustard (PMC), trans-4-phenylcyclophosphamide (T4P), didechlorocyclophosphamide (DCPA), acrolein (ACR), or ally1 alcohol (AA). Each bar is the mean + SEM for six mice expressed as a percentage of vehicle controls. The 50/50 DMSO/sahne control weights were (means k SD): ovary, 8.07 ? 1.36 mg; uterine weight/20 g body weight, 95.5 f 15.9 mg. The saline control weights were (means + SD): ovary, 8.42 + 10.97 mg; uterine weight/20 g body weight, 72.0 + 7.9 mg. *Statistically significant at p < 0.05 as determined by ANOVA and Duncan’s multiple range test. FIG.

volume by more than 50% and causing a significant destruction of all follicle types. Uterine weight was also significantly reduced by PMC and T4P. This effect was observed in the same dose groups that exhibited a significant destruction of antral follicles. The loss of these follicles can result in a significant depression in plasma estradiol concentrations (Plowchalk et al., 1989) and is probably the reason for the loss in uterine weight. It is possible that some of this uterine atrophy is the result of direct uterine toxicity, but when these compounds were administered by 101 which would minimize direct uterine effects, uterine weight loss was still observed. As seen with CPA, the primordial follicle pool was most sensitive and destruction was dose-dependent. Although the ED50 estimated for T4P is unreliable because of the limited number of data points on the curve, it is still apparent the dose-response curves of primordial follicle destruction for PM and T4P were significantly shifted to the left compared to CPA (Fig. 3). The data presented here indicates PMC and T4P are more potent than CPA on

OVARIAN

TOXICITY

479

an equimolar basis. In fact, all of the endpoints of ovarian toxicity measured were more severely affected by PMC and T4P than CPA. T4P and PMC even caused destruction of growing follicles, an effect not seen with CPA. The greater potency of T4P and PMC are probably due to differences in their metabolism compared to CPA. T4P undergoes metabolism very similar to CPA; however, the important urinary metabolites, 4-ketocyclophosphamide and carboxycyclophosphamide, are not formed. The presence of the phenyl group at the C-4 position prevents the oxidation reactions which typically generate these keto and carboxcylic acid metabolites. Because this eliminates two important detoxification pathways that account for approximately 25-40% of the dose excreted in the urine (Struck et al., 1971), a greater fraction of the parent compound will ultimately form PM. This may explain the greater ovarian toxicity of T4P compared to CPA. Direct administration of PMC also bypasses the bioactivation and detoxification steps normally required for the liberation of PM from CPA, therefore an equimolar dose of PMC is probably more toxic because it represents 100% conversion of CPA to PM. The compounds used to study the effects of ACR were DCPA and AA. The analog DCPA undergoes metabolism similar to CPA (Alarcon et al., 1972) but in the final elimination step releases ACR and phosphoric acid diamide, but not PM. Furthermore, the release of ACR from DCPA occurs at the same rate as it does from CPA (Wrabetz et al., 1980). AA is also known to generate ACR in vim, a reaction catalyzed by hepatic alcohol dehydrogenase (Serafini-Cessi, 1972), however, this enzyme is also found in the ovary (Holmes, 1978; Messiha, 1983). Neither DCPA nor AA had any effect on follicle numbers, ovarian volume, or uterine weight, even at the highest doses. DCPA, AA, and ACR did not produce ovarian toxicity even when directly injected into the ovary (101 studies). The inability of these compounds to produce any changes in

480

PLOWCHALK

the ovary suggest it is unlikely that CPA-generated ACR is involved in ovarian toxicity. In summary, these studies indicate that PM is the toxic metabolite responsible for CPAinduced ovarian toxicity. Only those compounds capable of generating PM caused ovarian damage similar to that seen with CPA. Furthermore, those compounds known to release ACR were ineffective as ovarian toxicants, indicating that ACR does not play a role in this toxicity.

REFERENCES ALARCON, R. A., MEIENHOFER, J., AND ATHERTON. E. (1972). Isophosphamide as a new acrolein-producing antineoplastic isomer of cyclophosphamide. Cancer Res. 32,5 19-523. BEAUCHAMP, R. 0.. ANDJELKOVICH, D. A.. KLINGERMAN, A. D., MORGAN, K. T.. AND HECK, H. A. (1985). A critical review of the literature on acrolein toxicity. CRC

Crit. Rev. Toxicol.

14, 309-380.

BOYD, V. L., ROBBINS.J. D.. EGAN, W., AND LUDEMAN, S. M. (1986). “P nuclear magnetic resonance spectroscopic observation of the intracellular transformation of oncostatic cyclophosphamide metabolites. J. Med. Chem. 29, 1206-1210. BROCK, N. (1967). Pharmacologic characterization of cyclophosphamide (NSC-2627 I) and cyclophosphamide metabolites. Cancer Chemother. Rep. 51, 3 15-325. BROCK, N. ( 1976). Comparative pharmacologic study in vitro and in vivo with cyclophosphamide (NSC-2627 I ), cyclophosphamide metabolites, and plain nitrogen mustard compounds. Cancer Treat. Rep. 60, 30 l-308. COLVIN, M., BRUNDRETT, R. B., KAN, M. N., JARDINE, I., AND FENSELAU. C. (1976). Alkylating properties of phosphoramide mustard. Cancer Res. 36, I 12 I - 1126. DELEAN, A., MUNSON, P. J., AND RODBARD, D. (1978). Simultaneous analysis of families of sigmoidal curves: Application to bioassay, radioligand assayand physiological dose response curves.Amer. J. Physiol. 235, E97E102.

(198 I b). Role of glutathione in the metabolism-dependent toxicity and chemotherapy of cyclophosphamide. Cuncer Res. 41. 3584-359 I. HOLMES. R. S. (1978). Electrophoretic analyses of alcohol dehydrogenase. aldehyde dehydrogenasc, aldehyde oxidase. sorbitol dehydrogenase and xanthine oxidase from mouse tissues. Comp. Biochem. Ph.wiol. 61. 339-46. JARRELL. J., YOUNG LIA. E. V., BARR, R.. MCMAHON. A.. BELBECK. L., AND O’CONNELL, G. (1987). Ovarian toxicity of cyclophosphamide alone and in combination with ovarian irradiation in the rat. Cancer Res. 47,23402343. KOYAMA, H.. WADA, T., NISHIZAWA, Y.. IWANAGA, T.. AOKI. Y., TERASAWA, T., KOSAIU, G., YAMAMOTO, T., AND WADA, A. (1977). Cyclophosphamide-induced ovarian failure and its therapeutic significance in patients with breast cancer. Cancer 39, 1403- 1409. LUDEMAN. S. M., BOYD, V. L., REGAN, J. B., GALLO, K. A.. ZON, G., AND ISHII, K. (1986). Synthesis of reactive metabolite analogues of cyclophosphamide for comparisons of NMR kinetic parameters and anticancer screening data. Drugs Exp. Clin. Res. 11, 527-532. MARINELLO, A. J., BERRIGAN, M. J., STRUCK. R. F., GUENGERICH, F. P., AND GURTOO, H. L. (1981). Inhibition of NADPH-cytochrome P450 reductase by cyclophosphamide and its metabolites. Biochem. Biophys. Res. Commun.

99, 399-406.

MATTISON. D. R., SHIROMIZU, K., AND NIGHTINGALE, M. S. ( I983a). Oocyte destruction by polycyclic aromatic hydrocarbons. In Reproductive To.xicology. (D. R. Mattison, Ed.), pp. 191-201. A. R. Liss, New York. MATTISON, D. R., SHIROMIZU, K., PENDERGRASS,J. A., AND THORGEIRSSON,S. S. (1983b). Ontogeny of ovarian glutathione and sensitivity to primordial oocyte destruction by cyclophosphamide. Pediatr. Pharmacol. 3, 49-55. MATTISON, D. R., THOMFORD, P. J., AND JELOVSEK,F. R. (1987). Effect of oocyte number and rate of atresia on the age of menopause. Reprod. Toxicol. 1,4l-5 I. MATTISON, D. R.. SINGH, I-I.. TAKIZAWA, K., AND THOMFORD, P. J. (1989). Ovarian toxicity of benzo(a)pyrene and metabolites in mice. Reprod. Toxicol. 3, 115-125. MEHTA. J. R.. PRZYBYLSKI, M., AND LUDLUM, D. B. (1980). Alkylation of guanosine and deoxyguanosine by phosphoramide mustard. Cancer Res. 40, 41834186.

GURTOO, H. C., DAHMS, R., HIPKENS, J., AND VAUGHT, J. B. (1978). Studies on the binding of [3H-chloroethyllcyclophosphamide and [ 14C4]cyclophosphamide to hepatic microsomes and native calf thymus DNA. Lfe

AND MATTISON

Sci. 22,45-52.

GURTOO, H. L., MARINELLO, A. J., STRUCK, R. F.. PAUL, B., AND DAHMS, R. P. (198 la). Studies on the mechanism of denaturation of cytochrome P-450 by cyclophosphamide and its metabolites. J. Biol. Chem. 256, 11,691-l

1.701.

GURTOO, H. L., HIPHENS. 1. H., AND SHARMA, S. D.

MESSIHA. F. S. (1983). Subcellular alcohol and aldehydedehydrogenases in the genital system of the female rat. Neurobehav.

Toxicol.

Teratol.

5, 247-250.

MILLER. J. J., AND COLE, L. J. (I 970). Changes in mouse ovaries after prolonged treatment with cyclophosphamide. Proc. Sot. E.up. Biol. Med. 133, 190-193. MIRKES, P. E. (1985). Cyclophosphamide teratogenesis: A review. Terato. Carcino. Mutagen. 5, 75-88. MIRKES, P. E., GREENAWAY, J. C., HILTON. J., AND BRUNDRETT, R. ( 1985). Morphological and biochemical aspects of monofunctional phosphoramide teratogenic-

CYCLOPHOSPHAMIDE ity in rat embryos cultured in vitro. Terutology32,24 I249. PEDERSEN,T., AND PETERS, H. (1968). Proposal for a classification of oocytesand follicles in the mouse ovary. .I. Reprod. Ferf. 17, 555-557. PLOTZ, P. H., KLIPPEL, J. H., DECKER, J. L., GRAUMAN, D., WOLFF, B., BROWN, B. C., AND RUIT, G. (1979). Bladder complications in patients receiving cyclophosphamide for systemic lupus erythmatosus or rheumatoid arthritis. Ann. Intern. Med. 91, 221-223. PLOWCHALK, D. R., AND MATTISON, D. R. (1989a). Ovarian morphometric changes following cyclophosphamide treatment. Growth Factors and The Ovary. (A. N. Hinhfield, Ed.), pp. 427-432. Plenum, New York. PLOWCHALK, D. R., AND MATTISON, D. R. (1989b). The role of altered hepatic glutathione levels in reduced ovarian toxicity with fractionated doses of cyclophosphamide. The Toxicologist 9, 30. [Abstract] PLOWCHALK, D. R., MEADOWS, M. J., AND MATTISON, D. R. (1989). Changes in the uterotropic response to estradiol following cyclophosphamide treatment. Amer. Coil. Toxicol. Fall 1989. [Abstract] SAS User’s Guide: Statistics. (1982). Statistical Analysis System Institute, Inc., Gary, North Carolina. SERAFINI-CESSI,FRANCA. (1972). Conversion of ally1 alcohol into acrolein by rat liver. Biochem. J. 128, 11031107. SHIROMIZU, K., ‘THORGEIRSSON,S. S., AND MATTISON, D. R. (1984). Effect of cyclophosphamide on oocyte and follicle number in Sprague-Dawley Rats, C57BL/ 6N and DBA/2N mice. Pediutr. Pharmacol. 4, 2 13221.

OVARIAN STRUCK,

481

TOXICITY

R. F., KIRK,

M. C., MELLETT,

L. B., EL DAREER,

S., AND HILL, D. L. ( 197 1). Urinary metabolites of the antitumor agent cyclophosphamide. Mol. Phurmacol. 7,5 19-529. TAKIZAWA, K., YAGI, H., JERINA,D. M., AND MATTI.SON, D. R. (1984). Murine strain differences in ovotoxicity following intraovarian injection with benzo(a)pyrene, (+)-(7R,SS)-oxide, (-)-(7R,IR)-dihydrodiol, or (+)(7R,SS)diol-(9S, lOR)-epoxide-2. Cancer Res. 44,257 l2576. WARNE, C. L., FAIRLEY, K. F., HOBBS,J. B., AND MARTIN, F. R. (1973). Cyclophosphamide-induced ovarian failure. N. Engl. J. Med. 289, 1159-I 162. WILDNAUER, D. B., AND OEHLMANN, C. E. (1982). Interaction of cyclophosphamide metabolites with membrane proteins: An in vitro study with rabbit liver microsomes and human red blood cells. Effects of thiols. Biochem. Phurmacol. 31,3535-3541. WILMER, J. L., EREXSON, G. L., AND KLIGERMAN, A. D. (1986). Attenuation of cytogenetic damage by 2-mercaptoethanesulfonate in cultured human lymphocites exposed to cyclophosphamide and its reactive metab olites. Cancer Res. 46, 203-210. WINCKLER,

K., OBE. G., MADLE,

S., AND

NAU,

H. (1984).

Mutagenic activities of cyclophosphamide (NSC-26271) and its main metabolites in Salmonella typhimurium. human peripheral lymphocytes and Chinese hamster ovary cells. Mutat. Res. 129,47-55. WRABETZ, E.. PETER, G.. AND HOHORST, H. J. (1980). Does acrolein contribute to the cytotoxicity of cyclophosphamide? J. Cancer Res. Clin. Oncol. 98, 119- 126.