Phase I Trial of the Thymidylate Synthase Inhibitor …clincancerres.aacrjournals.org/content/clincanres/2/10/...Vol. 2, 1685-1692, October 1996 Clinical Cancer Research 1685 Phase

Vol. 2, 1685-1692, October 1996 Clinical Cancer Research 1685

Phase I Trial of the Thymidylate Synthase Inhibitor AG331 as a

5-Day Continuous Infusion’

Peter J. O’Dwyer,2 Paul B. Laub,

Deborah DeMaria, Mingxin Qian, Denise Reilly,

Bruce Giantonio, Amanda L. Johnston,

Ellen Y. Wu, Lisa Bauman, Neil J. Clendeninn,

and James M. Gallo

Fox Chase Cancer Center, Philadelphia, Pennsylvania 1911 1 [P. J. 0.,

P. B. L., D. D., M. Q., D. R., B. G., J. M. G.], and AgouronPharmaceuticals, La Jolla, California 92121 [A. L. J., E. Y. W., L. B.,

N.J.C.]

ABSTRACT

AG331 (N�-[4-(morpholinosulfonyl)benzyl]-N’�-methyl-2,6-diaminobenz-[c,d}-indole glucuronate) is a lipophilic thy-

midylate synthase inhibitor with activity in solid tumor

models. On the basis of preclinical data supporting regimens

of frequent drug administration, we performed a Phase I

trial of AG331 as a 5-day continuous infusion repeated every

3 weeks. Twenty-nine patients were entered at doses ranging

from 25 to 1000 mg/m2/day. The major side effects were

mild to moderate fatigue, nausea, vomiting, diarrhea, andfever. At doses �400 mg/m2, acute reversible elevation of

bilirubin, aspartate aminotransferase, alamne aminotrans-

ferase, and �y-glutamyltranspeptidase was observed. All pa-

tients who received �600 mg/m2/day experienced elevated

alanine aminotransferase. Elevated liver function tests were

evident by day 3 of the infusion and had resolved by day 8

in the majority. This toxicity was dose limiting at 1000

mg/m2/day, at which dose two of two patients developedgrade 4 reversible hyperbilirubinemia in addition to the

enzyme elevations. Serum and urine samples were analyzed

by a novel high-pressure liquid chromatography method for

the determination of the pharmacokinetics of AG331. Overthe 50-1000 mglm2/day dose range, mean total clearance

ranged from 11.6 to 30.0 liters/hIm2, and volume of distri-

bution at steady state ranged from 279.5 to 758.7 liters/m2.

These parameters were dose independent over the dose

range tested. The harmonic mean terminal half-life of

AG331 was 20.2 h Less than 5% of an AG331 dose is

eliminated unchanged in the urine. Both the administered

dose and exposure to the drug were related to the changes in

bilirubin and aminotransferase blood levels. Evidence for

inhibition of thymidylate synthase was obtained at doses

ranging from 100 to 1000 mg/m2 in seven patients; plasma

deoxyuridine concentrations at end-infusion were 1.8-3.8-

fold higher than pretreatment values. Because of the natureof toxicity on this schedule, more extensive Phase II evalu-

ation is not recommended, although an AG331 dose of 800mg/m2/day for 5 days is tolerable. Exploration of less fre-

quent dose administration is under way.

INTRODUCTION

Inhibition of thymidylate synthase, a key enzyme in the de

novo synthesis of thymidine nucleotides, results in the arrest of

DNA synthesis and cytotoxicity in many human cell types (1).

The most widely used inhibitors of thymidylate synthase are the

fluorinated pyrimidines; the anabolite FdUMP3 is a potent in-

hibitor (2). However, despite their substantial therapeutic activ-

ity, particularly in tumors of gastrointestinal or breast origin, the

fluorinated pyrimidines exert toxic effects at numerous other

biochemical boci (3). Toxicity has been correlated with incor-

poration of fluorinated nucleotides into both RNA and DNA (4,

5); these effects have been held to contribute more to normal

tissue toxicity than to antitumor activity (6). Furthermore, the

potency of FdUMP varies with the availability of reduced folate;

cells deficient in folate are relatively resistant to inhibition (7).

These characteristics of fluoropyrimidines prompted the devel-

opment of more potent and more selective inhibitors of thymi-

dylate synthase.

The design of such inhibitors was made possible by the

determination of the three-dimensional structure of the active

site of thymidylate synthase (8). X-ray crystallography revealed

the characteristics of this structure, as well as the binding of

ligands to it. A number of candidate inhibitors were then devel-

oped through de novo design at the reduced folate binding site

on the protein. The molecules were optimized by iterative se-

quences of cocrystallization, protein-ligand structure determina-

tion, and binding studies. AG33 1 (Fig. 1 ) is a potent inhibitor of

thymidylate synthase that has emerged from this rational syn-

thesis program (9).

AG33 1 inhibits thymidylate synthase with a K� for the

human enzyme of 0.4 nM. IC50s toward a range of human and

murine cell lines are in the low micromolar range ( 10). In the

National Cancer Institute in vitro human tumor screen, AG33 I

was selectively toxic to colon cancer and leukemia cell lines. In

vivo, preclinical antitumor activity was identified in L5 178Y/

TK- lymphoma and the human colon cancer xenograft GC1IM/

(TK-). Schedules of frequent and protracted (10-day) treatment

were associated with superior activity (10).

Received 10/13/95; revised 6/12/96; accepted 6/17/96.

I Supported by Grant CA 06972 from the National Cancer Institute and

an appropriation from the Commonwealth of Pennsylvania.2 To whom requests for reprints should be addressed, Thomas JeffersonUniversity, Bluemle Life Sciences Building, 233 South 10th Street,

Suite 502, Philadelphia, PA 19107.

3 The abbreviations used are: FdUMP, 5-fluorodeoxyuridine 5-mono-phosphate; AG33 I , N�-[4-morpholinosulfonyl)benzyll-M�-methyl-2,6-

diaminobenz-[c,d]-indole glucuronate; AST, aspartate aminotransferase:ALT, alanine aminotransferase; HPLC, high-pressure liquid chromatog-raphy; AUC, area under the curve; dUrd, deoxyuridine; 5-FU, 5-flu-

orouracil.

Research. on January 28, 2020. © 1996 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

S

1686 Phase I Trial of AG331

Fig. 1 Structure of AG331.

The pharmacokinetics and tissue distribution of AG331

have been described in mice, rats, and dogs (1 1). The terminal

half-life in the mouse ranges from 2 to 8 h, compared with 3.8 h

in the rat and a mean of 13. 1 h in the dog. No evidence for

dose-dependent kinetics was obtained in any of the species. The

toxicology studies demonstrated a single dose LD10 of 145

mg/kg (435 mg/m2) in mice. A single dose of 60 mg/kg (1200

mg/m2) was tolerated by beagles with acute reversible symp-

toms of histamine release, hypoactivity, and emesis. In a repeat

dose (days 1 , 3, and 5) study in dogs, the 45 mg/kg (900 mg/m2)

dose level was tolerated, with similar symptoms. At this dose,

reversible elevations in serum alkaline phosphatase, ALT, and

AST were observed. Reddening and swelling at the injection site

were common.

On the basis of the data favoring frequent and prolonged

exposure to AG331, we performed a Phase I trial of AG331 on

a 5-day continuous infusion schedule. This trial incorporated

pharmacokinetic and pharmacodynamic analyses and deter-

mined a maximum tolerated dose on this schedule.

MATERIALS AND METHODS

Patient Population

Patients eligible for this study had a histological diagnosis

of a malignant solid tumor and had exhausted the standard

therapeutic options for their disease or had a malignant disease

for which no established therapy exists. They were 18 years of

age or older and had an Eastern Cooperative Oncology Group

performance status of 0 or 1 . They had adequate bone marrow

(absolute granulocyte count, � 1,500/mm3; platelet count,

� 100,000/mm3), liver function (bilirubin, < 1.5 mg/dl; AST or

ALT, <5-fold the institutional upper limit of normal), and

kidney function (creatinine, < 1.5 mg/dl). Patients had recovered

from all toxicities of prior treatment and had no prior chemo-

therapy or radiotherapy within 4 weeks of entry to this study (6

weeks for drugs with delayed toxicity, such as nitrosoureas or

mitomycin). All patients gave written informed consent in ac-

cordance with federal, state, and institutional guidelines.

Before therapy, a medical history, physical examination,

complete blood count, biochemical profile, electrocardiogram,

urinalysis, and chest X-ray were performed. Patients were mon-

itored with complete blood counts twice a week and biochem-

ical profiles once a week. Physical examination, as well as

X-rays and scans as required for tumor measurement, were

performed before each course to assess response. When com-

puted tomography scans or magnetic resonance imaging was

required, these tests were performed every other course.

Individual patients did not have drug doses escalated in

subsequent chemotherapy cycles. Dose modifications were not

made for nausea and vomiting, alopecia, anemia, or venous

irritation. The dose for subsequent cycles of treatment was

determined by the toxicity experienced in the first course. Pa-

tients in whom grade 3 toxicity developed had a 25% dose

reduction for subsequent cycles, and those in whom grade 4

toxicity developed had a 50% dose reduction. Persistent toxicity

on the day of planned treatment led to a delay in treatment until

resolution of symptoms.

Results are reported using a modification of the consensus

toxicity criteria (Cancer Therapy Evaluation Program, National

Cancer Institute, Bethesda, MD; 1988). Response criteria were

standard (12). The maximum tolerated dose of AG33 I was

defined as the dose that would produce a >33% incidence of

grade 4 myelosuppression or a >33% incidence of grade 3

nonmyebosuppressive toxicity.

Treatment Plan

Patients were admitted to the Mary S. Schinagl Clinical

Studies Unit at Fox Chase Cancer Center for the initial course of

AG33 1 . AG331 was provided by Agouron Pharmaceuticals

(San Diego, CA) as orange-red lyophilized powder for injection

in 5-ml clear vials containing 60 mg of AG331 and 75 mg of

mannitol. Reconstitution of the material was accomplished by

dilution in 5% dextrose to a total volume of 500 ml for each

24-h period. The solution was administered as a continuous iv.

infusion through a central line (in most patients, an in-dwelling

port) over S days.

The starting dose of AG33 I was 25 mg/m2/day, a value

that was 1/32 of the maximum tolerated dose in a 5-day dog

study and one that had been tolerated in a parallel single-dose

study. The strategy for dose escalation was to double the dose

until any evidence of toxicity was observed, at which time

escalation was to proceed in 25-50% increments. Provision was

made to expand accrual to a level on encountering severe or

unexpected toxicity. The end point of the study was to describe

a dose of AG33 I at which fewer than one-third of the patients

would experience grade 3 or grade 4 toxicity and, therefore, to

define a regimen suitable for a broad Phase II testing.

Pharmacokinetic Studies

The pharmacokinetics of AG33 1 were determined in 25 of

the 29 patients entered in this study. Blood samples were drawn

into Vacutainer tubes, allowed to clot at room temperature, then

placed on ice and centrifuged at 4#{176}Cfor 5 mm at 2500 X g.

Serum was stored at -20#{176}Cuntil analysis. Samples were ob-

tamed before treatment and at I , 2, 12, 24, 48, 72, 96, and 120 h

after the initiation of the continuous infusion. After discontinu-

ation of the infusion, samples were obtained at 15, 30, and 45

mm and at 1, 2, 4, 6, 8, 12, 18, 24, 36, and 48 h. Blood collection

times were recorded from the start of the 5-day drug infusion.



Table 1 Characteristics of treated patients

No. of patients entered 29

Males/Females 21/8

Median age (range) 58 (38-77)

Performance status0I

53

Prior therapy

None 1

Chemotherapy 14

Chemotherapy/radiation 14

Primary tumor

Colorectal S

Esophageal 2Pancreas 2

Head and neck 2

Other 8

Clinical Cancer Research 1687

HPLC Analysis of AG331

AG33 1 concentrations in serum and urine samples were

quantitated by a HPLC method published previously (13).

Briefly, preconditioned Bond Elut C18 cartridges (3 ml) were

loaded with either I ml of serum or urine. Each cartridge was

washed with 3 ml of Dulbecco’s PBS two times, followed by

elution of AG33 1 with 3 ml of methanol. The methanol fraction

was dried at 40#{176}Cunder N, gas. Residues were reconstituted in

200 �i.l of mobile phase that consisted of 35:65 (v/v) acetonitrile:

water containing 25 mrvi ammonium phosphate and phosphoric

acid to adjust pH to 3.5. Reconstituted samples were transferred

and centrifuged before injection onto a HPLC system. AG33I

was detected at 457 nm following separation on a C18 column

(250 X 4.6 mm inside diameter) using a mobile phase flow rate

of 1 ml/min. Intra- and interday accuracy and precision of

AG33 1 analyses were routinely � 15%, with a limit of quanti-

tation of 20 ng/ml.

Noncompartmental Pharmacokinetic Analysis

Pharmacokinetic parameters were obtained by noncom-

partmental analysis of plasma AG33 1 concentrations. AUC and

its associated first moment curve were computed by Lagrange

polynomial interpolation and integration from time 0 to the last

measured time. Extrapolation of the area from that point to time

infinity was made through estimation of the terminal rate con-

stant using unweighted least squares fitting to the terminal slope

of the natural logarithm of concentration versus time. The phar-

macokinetic analysis used the computer program NCOMP�,

which is based on the LAGRAN program (14). Both mean

residence time (MRT) and volume of distribution at steady state

(V��) were corrected for the duration (7) of the infusion. The

equations for the parameters are as follows (15, 16):

DoseCL =

AUMC TMRT= AUC �

Dose X AUMC Dose X Tvs.s = AUC2 � 2 X AUC

The mean terminal half-life (T112) was computed as the har-

monic mean, and the pseudo-SD was computed by the jack-

knife method (17).

Pharmacodynamic Analysis

Correlation between measures of hepatotoxicity, elevation

in grade of bilirubin level, percentage elevation in AST and

ALT concentrations, AG331 dose, AUC, and end of infusion

concentration were analyzed by computing Pearson’s correla-

tion coefficient.

4 P. B. Laub, and J. M. Gallo. NCOMP: a Windows-based computerprogram for noncompartmental analysis of pharmacokinetic data, sub-

mitted for publication.

Circulating Deoxyuridine Concentrations

The pharmacodynamics of AG331 were also studied by

measuring the effects of the drug on the plasma or serum

concentration of 2’-dUrd in selected patients. The assay, mod-

ified after the techniques of Rafi et a!. ( 1 8), was performed in

two stages and is described briefly below.

Sample Preparation. Fifty �il of [3H]dUrd (7 p.Ci/ml,

22 Ci/mmol; Moravek Biochemicals, Inc., Brea, CA) were

added as the internal standard to samples of thawed patient

plasma (0.5-1.0 ml), which were then extracted with acetoni-

trile. The contents of the tubes were then centrifuged for I 0 mm

at 1000 rpm at 4#{176}C,and the supernatants were dried at 40#{176}C

under a stream of nitrogen.

Initial Separation. Samples were resuspended in 200 �i.l

of the initial separation HPLC mobile phase and analyzed on a

Beckman System Gold model 338 HPLC system (Beckman

Instruments, Fullerton, CA) by injection via a Shimadzu model

SIL-9A autosampler (Shimadzu Corp., Kyoto, Japan) onto a C18

column (Jones Chromatography Nucleosil ODS, 3 p.m, 100 X

4.6 mm) fitted with a self-packed guard column (Whatman

LRP-2 ODS, 37-53 p.m, 20 X 2 mm). The mobile phase was

94.5% ammonium acetate [0.05 M (pH 5)]:5.5% methanol, v/v.

The elution conditions were isocratic, and the flow rate was 1

ml/min. The effluent was monitored for radioactivity, and frac-

tions were collected automatically with a Gibson model FC-204

fraction collector (Gibson Medical Electronics, Middleton, WI)

under the control of a Beckman model 17 1 radioisotope detector

(Beckman Instruments). Under these conditions, dUrd had a

retention of -4.5 mm.

Second Separation. Dried samples from the initial sep-

aration were reconstituted in the second separation mobile phase

and analyzed for dUrd using the above HPLC system and a C6

column (Jones Chromatography Spherisorb, 3 p.m, 100 X 4.6

mm). The mobile phase was 98.5% formic acid [0.2%, (pH

3.25)]:l.5% acetonitrile, v/v. The elution conditions were iso-

cratic, and the flow rate was 1 mb/mm. The effluent was mon-

itored for radioactivity by a Beckman model 171 radioisotope

detector (Beckman Instruments) and for UV absorption (262

and 280 nm) with a Gilson model 1 19 UVIVIS detector (Gibson

Medical Electronics). dUrd was identified by retention time,



0 1 2 3day4 5 6 7

350

�.300

�. 250

�200

,� 150

�1000

.� 50E(5 0

300

-I 250

�, 200

C 50Eca

-j

(5UI(5

(5

UIC15

0C

E15

0 1 2 3day4 5 6 7

0 1 2 3day4 5 6 7

1688 Phase I Trial of AG331

Table 2 Eleva tions in li ver functi on test s in th e first cycle

Dose Points

Bilirubin AST ALT

1 2 3 4 1 2 3 4 1 2 3 4

600

800

1000

662

32 2

2

21

351

1

1

421

13

111

radiochemical detection, and the 262:280-nm wavelength ratio.

The final plasma dUrd concentration was corrected for recovery

based on the radiochemical content of the [3H]dUrd relative to

that initially added.

RESULTS

Twenty-nine patients were entered and received 55 courses

of AG331. Doses were escalated from 25 mg/m2/day to 1000

mg/m2/day. The drug was administered as a 5-day continuous

infusion. The demographic characteristics of the patients entered

on the study are shown (Table 1). The patients were of excellent

performance status (median performance status, 1), and all but

one patient had received prior chemotherapy. One patient was

found on subsequent review to have a performance status of 2.

Of the 28 patients who received prior chemotherapy, 23 had

prior exposure to 5-FU. The distribution of tumor types reflected

the activity of 5-EU, and the majority of tumors (75%) was of

gastrointestinal origin.

Clinical Toxicity. The major and dose-limiting toxicity

of AG331 by the 5-day infusion schedule was hepatotoxicity. At

doses �600 mg/m2/day, all patients experienced acute reversi-

ble elevations in bilirubin, AST, and ALT (Table 2). In the first

two patients who experienced these toxicities, the infusion was

terminated early, and values returned to normal within 24 h. The

infusion was not interrupted for subsequent patients, and the

toxicities did not change in severity or reversibility. As shown in

Fig. 2, aminotransferase values generally peaked on day 3 or 4,

remained elevated until the infusion was discontinued, then

returned to baseline usually by days 8-10. In patients who

received more than one cycle (maximum 4), there was no

evidence of cumulative hepatotoxicity. Two patients treated at

1000 mg/m2/day were both discontinued after 4 days of treat-

ment as a result of severe (grade 4) hyperbilirubinemia (3.9 and

5.5 mg/dl, respectively).

One of the patients at the highest AG331 dose was a

72-year-old male with rectal cancer metastatic to lung. In addi-

tion to hyperbilirubinemia, he experienced nausea, vomiting,

fatigue, and fevers during the course of the infusion. After

discontinuation of the AG331 on day 4, other symptoms re-

solved but fever persisted. In the subsequent days, he developed

bilateral parenchymal lung infiltrates. A transbronchial biopsy

was remarkable for proteinaceous intraalveolar exudates and

mild arteritis. No tumor cells or organisms were seen. The

patient was treated with antibiotics (trimethoprin/sulfamethox-

azole), and the infiltrates resolved. Symptoms, however, re-

mained over a 3-month period after treatment, during which his

metastatic lung disease progressed. The relationship of the pul-

monary infiltrates to AG331 is unclear.

Other nonmyeloid effects were anorexia, nausea, vomiting,

Fig. 2 Profile of elevation in aminotransferases in individual patientstreated at each of the three highest dose levels. Top, 600 mg/m2/day;

middle, 800 mg/m2/day; bottom, 1000 mg/m2/day.

fever, and fatigue. At doses �600 mg/m2/day (14 patients total),

fatigue was a significant toxicity (grade 2 in seven patients).

Nausea and vomiting of grade 2 or greater were observed in five

patients (36%). Prochlorperazine-based regimens were effective

antiemetic therapy. Myelosuppression was strikingly absent in

all patients following AG331 administration. Anemia was gen-

erally mild, without an apparent relationship to dose or to drug

administration. Five patients manifested grade 3 anemia; how-

ever, all of these patients had preexisting grade 1-2 anemia.



3000

2500

E�. 2000C)C

C.2 1500CS

C

� 1000C00

500

25 50 75 100 125 150 175 200

hours


Fig. 3 Mean AG331 concentration (bars, ± 1

SD) for the six subjects at the 800-mg/m2/day

dose levels.

Table 3 Pharmacokinetic parameters of AG33 I administered by 5-day infusion

Dose

(mg/m2) N

AUC (m

Mean

glhlliter)

SD

CL (lite

Mean

rfh/m2)

SD

V� (lit

Mean

er/m2)

SD

MR

Mean

I (h)

SD

So 2 15.2 8.7 19.7 11.3 654.0 441.0 32.1 4.0

100 3 32.7 27.2 23.6 16.1 759.0 225.0 45.2 33.6

200 3 44.5 14.5 24.3 8.8 627.0 429.0 24.2 8.2

400 3 65.5 37.3 30.0 13.2 372.0 100.0 13.3 3.5

600 6 228.0 1 19.0 14.4 4.9 280.0 144.0 19.8 5.8800 6 214.0 45.8 19.6 4.9 408.0 117.0 21.7 7.1

1000 2 550.0 360.0 1 1.6 7.6 524.0 342.0 45.3 0.1

25 20.0 9.5 470.1 261.4 26.1 15.3�1 Harmonic mean.b Pseudo-SD.

The first two patients at the lowest dose bevel were treated

using peripheral venous access; both developed phlebitis within

24 h. All subsequent patients were treated through central ye-

nous access. Three patients, one with a femoral line and two

with subclayian ports, developed thrombotic episodes that re-

quired anticoagulation. The thrombosis was local in two pa-

tients, and in one, recurrent lower extremity deep venous throm-

boses accompanied drug dosing. There have been no objective

responses with AG33 1 administered as a 5-day continuous

infusion.

Pharmacokinetics. An average of I 9 concentration

measurements (range, 12-22) was obtained for each of 25

subjects. In two of the subjects, infusion was terminated before

120 h at 69.5 and 84 h. The mean plasma concentration-time

profile at the 800 mg/m2/day dose level is presented in Fig. 3.

Pharmacokinetic parameters for AG33 1 computed from all

25 subjects are shown in Table 3. The overall CL and volume of

distribution at steady state (V�) were 20.0 ± 9.5 liters/hIm2 and

470.1 ± 261.4 liters/rn2, respectively. The harmonic mean half-

life was 20.2 ± 15.9 h; thus, steady-state AG331 concentrations

would be achieved during the last day of treatment. Analysis of

AG331 in urine in a small number of subjects indicated that

<5% of the dose was excreted as unchanged drug.

The relation of CL and � to the administered AG33 1

dose is illustrated in Fig. 4. CL and � are independent of

dose over the range measured as determined by a one-factor

ANOVA (P > 0.05) as well as by the nonparametric Kruskal-

Wallis test (P > 0.05; Ref. 19). There is a trend toward a

lowered CL and V,� with increasing dose, but there is also

substantial interpatient variability that may mask saturable

processes, suggested by consideration of the bower dose

levels (20). No patient had pharmacokinetic analyses per-

formed at more than one dose.

Pharmacodynamics. Analyses of the interactions be-

tween pharmacological determinants and drug effects showed

that dose correlates most strongly with measures of hepatotox-

icity, having Pearson correlation coefficients (R) of 0.68 with

grade of bilirubin elevation, 0.80 with grade of AST elevation,

and 0.65 with peak ALT level. AUC correlates more weakly

with these measures, with Rs of 0.57, 0.61, and 0.35, respec-

tively. Likewise, end-of-infusion concentration resulted in Rs of

0.55, 0.67, and 0.41, respectively.

The pharmacodynamics of AG331 were evaluated further

by the analysis of serum or plasma dUrd concentrations. dUrd is

derived from intracellular dUMP bevels, which increase after

thymidylate synthase inhibition (2 1-23). Therefore, increased



A

B

.

.

.

. .. .

I . IIS

oJ

E

-I

U)U)

>

1200

1000

800

600

400

200

0

0 200 400 600 800 1000

dose(mg/m2/day)

S

SS

S

S SS S

S � #{149} f

0 200 400 600 800 1000

dose (mg I m2 I day)

1690 Phase I Trial of AG33 1

45

40__35c’JE �

� 25

d. 20-J0 15

10

5

0

Fig. 4 Total clearance (A) and volume of distribution at steady state(B) as a function of AG331 dose.

plasma or serum dUrd levels in patients may reflect thymidylate

synthase inhibition in vivo (18, 24). Serum or plasma dUrd

levels were measured in seven patients treated with one course

of AG331 in doses ranging from 100 to 1000 mg/m2/day. The

dUrd concentrations were measured before AG33 1 administra-

tion, at the end of the 5-day continuous infusion, and 24 h after

the end of infusion. There was considerable interpatient varia-

bility in the preinfusion plasma dUrd levels, ranging from 0.020

to 0.1 15 p.M dUrd. dUrd bevels were elevated in all patients at

the end of the I 20-h continuous infusion; plasma dUrd concen-

trations were 1.3-3.8-fold higher than pretreatment levels (Ta-

ble 4). However, at 24 h after the end of infusion, all dUrd levels

had returned close to pretreatment concentrations.

DISCUSSION

Thymidylate synthase, the enzyme required to synthesize

the only nucleotide unique to DNA, was the target selected by

Heidelberger (25) in his synthesis of 5-FU as a cytotoxic drug.

Santi et a!. (2) elucidated the biochemical mechanism of the

enzyme and the interaction of FdUMP in its inhibition. The

requirement for reduced folate in forming the ternary complex

of nucleotide-folate-enzyme led to the prediction that replenish-

ment of folate pools would maximize the potential for complex

formation and so enhance the potency of 5-FU. Berger and

Hakala (26) verified this in vitro, and subsequent clinical trials

demonstrated that the addition of leucovorin increased the re-

sponse rate of 5-FU in colorectal cancer (27). These observa-

tions gave rise to a number of approaches to the synthesis of

more potent and more specific thymidylate synthase inhibitors.

Most efforts have targeted the folate binding site on the

enzyme. An early empirical derivative was the N’#{176}-propargyl

folate analogue CB3717 (2], 28). Clinical trials of this com-

pound were halted when nephrotoxicity proved dose limiting in

Phase I (29). This effect resulted from precipitation in the

tubules of the kidney, a consequence of the poor sobubility of the

drug. It also may be noted that this compound was hepatotoxic,

through a mechanism that remains poorly understood.

AG33 1 , on the other hand, was developed from a program

of computer-based chemical synthesis, modeled from a knowl-

edge of the three-dimensional crystallographic structure of the

active site of human thymidylate synthase (1). Another of this

series also is in clinical trials; AG-337 has toxicity expected of

a thymidybate synthase inhibitor, with myebosuppression and

gastrointestinal toxicity predominating (30).

AG33 1 is a lipophilic-specific inhibitor of thymidylate

synthase with a low K,, paralleled by its potent inhibition of

cultured cancer cell lines (10). The specificity of this effect is

supported by its ready reversal with exogenous thymidine.

AG331 is not, however, a substrate for polyglutamylation. Thus,

it was hypothesized that continuous exposure would maximize

its therapeutic activity in human tumors. This hypothesis was

supported by preclinical data that promised superior activity

with regimens of frequent administration. Hence, a continuous-

infusion, 5-day regimen was piloted in this Phase I study.

Unexpectedly, however, hepatic toxicity supervened at the

higher doses. This toxicity was dose dependent and proved dose

limiting at 1000 mg/m2/day. Although brief in duration and

without evident bong-term sequelae, this toxicity may not be a

consequence of thymidylate synthase inhibition. Hepatocytes

have an extremely low proliferative index and an excellent

capacity to salvage thymidine from the circulation. The appear-

ance of this effect in the absence of myelosuppression or gas-

trointestinal toxicity suggests that, by this schedule, inhibition of

another target is responsible. Coupled with the lack of responses

(in striking contrast to AG-337), this schedule is not recom-

mended for further development in Phase II trials. However,

using the information available from this study, alternative

schedules are currently being piloted.

The other toxic effects observed in this trial (fatigue, leth-

argy, and headache) are shared to a greater or lesser degree by

other specific thymidybate synthase inhibitors, including Tomu-

dex (Dl694) and LY231514 (31, 32). The mechanism of these

effects is also unclear, and a contribution from altered hepatic

function is possible. In addition, the preclinical toxicological

studies suggest that, like Trimetrexate, AG33 1 may inhibit

histamine N-methyltransferase. However, other than headache,

no acute evidence of such inhibition has been observed in

humans.

The clinical pharmacokinetics presented here show that in

humans, as in the preclinical models, AG331 is cleared largely

by nonrenal mechanisms. It may be anticipated that the hepa-

tobiliary route of elimination is a component of the nonrenal

pathways, although direct evidence for this was not obtained in




5 J. M. Gallo, unpublished results.

Table 4 Circulating 2’-d Urd levels (p.M) in patients tre ated with AG33I

Patient (dose) Source Preinfusion

Postinfusion

30 mm” 24 h

KF (100 mg/m2) Serum 0.049 0.062 (1.3) 0.060

ST (400 mg/m2) Serum 0.020 0.076 (3.8)” 0.024BH (600 mg/m2) Serum 0.038 0.126 (3.3) 0.041CP (600 mg/m2) Plasma 0.043 0.066 (1.5) 0.027JY (600 mg/m2) Serum 0.1 14 0.160(1.4)” 0.059

JG (800 mg/m2) Serum 0.1 15 0.264 (2.3)” 0.1 15DM (1000 mg/m2) Serum 0.106 0.261 (2.5)� 0.218

a Values in parentheses represent fold increase over preinfusion.

6 Sample taken 15 mm postinfusion end.

C Sample taken 96 h into infusion. Postinfusion, 15 or 30 mm. Sample not available.

the current study. Clearance of AG33 1 can be characterized as

moderately efficient (mean clearance of AG33 1 , 20 liters/h/m2

versus human liver blood flow of 90 biters/h). Apparent AG33 I

metabolites, most likely glucuronides, have been identified in

the bile of rats, and our preliminary data suggest that at the

higher doses, these also may be identifiable in human plasma.5

The high volume of distribution values indicates extensive

tissue uptake. This may be a result both of protein binding and

of partitioning into membranes and other lipid structures. Over-

all, AG331 kinetics can be characterized as linear (Table 3), but

interpatient variability precludes a definite statement concerning

dose-dependent kinetics, and studies directed to this issue are

warranted. Finally, the long elimination half-life of AG33 1

makes long-term infusional regimens unnecessary.

Serum or plasma dUrd concentrations were analyzed as an

indirect measure of thymidylate synthase inhibition. The limited

measures of serum and plasma dUrd from this study suggest that

some inhibition of the enzyme is indeed achieved by AG33 1,

although the data do not lend themselves to statistical analysis.

The measured increases in plasma dUrd levels are similar to

those observed with AG-337 (18) and show the same lack of

dose dependency (Table 4), although such correlations would be

best examined using multiple dUrd measurements in each pa-

tient. Similar to results reported by Rafi et a!. (1 8) in studies

with AG-337, the magnitude of the dUrd increase does not

increase with increasing AG33 1 dose, suggesting that maximal

cellular export of dUrd may be achieved at doses as bow as I 00

or 400 mg/m2/day AG331 . After discontinuation of the infusion,

recovery occurs within 24 h, consistent with reversible inhibi-

tion of thymidylate synthase by AG33 I , and with relatively

rapid clearance of the nonpolyglutamated AG33 1 from intraceb-

lular stores. However, definitive interpretation of these data

would benefit from simultaneous tissue thymidybate synthase

measurements.

Consideration of the clinical and pharmacological data,

therefore, prompts a revised approach to AG33 1 administration.

The prolonged half-life of the drug makes bong-term infusions

unnecessary. The occurrence of hepatic toxicity might also

rebate to this schedule, possibly by permitting the accumulation

of a toxic metabobite. The infusion alone, however, is not

responsible; similar hepatotoxicity was observed in the parallel

5-day bobus schedule.6 On the basis of the findings in this initial

Phase I trial, intermittent schedules will be explored and tested

in Phase II studies.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the expert secretarial assist-

ance of Catherine Thompson.

REFERENCES

I . Jackson, R. C., Boritzki, T. J., Johnston, A. L., Shetty, B. V., Varney,

M. D., Webber, S., and Webber, S. E. Design and development oflipophilic inhibitors of thymidylate synthase. Proc. Am. Assoc. Cancer

Res., 33: 592, 1992.

2. Santi, D. V., and McHenry, C. S. 5-Fluoro-2’-deoxyuridylate: cova-

lent complex with thymidylate synthetase. Proc. Natl. Acad. Sci. USA,

69: 1855-1857, 1972.

3. O’Dwyer, P. J. Biochemical modulation as an approach to the rever-sal of antimetabolite resistance. In: L. J. Goldstein and R. F. Ozols(eds.), Anticancer Drug Resistance: Advances in Molecular and Clinical

Research, pp. 201-216. New York: Kluwer Academic, 1994.

4. Kufe, D., and Major, P. 5-Fluorouracil incorporation into humanbreast carcinoma RNA correlates with cytotoxicity. J. Biol. Chem., 256:9802-9805, 1981.

5. Herrick, D. J., Major, P. P., and Kufe, D. W. Effect of methotrexateon incorporation and excision of 5-fluorouracil residues in human breastcarcinoma DNA. Cancer Res., 42: 5015-5017, 1982.

6. Houghton, J. A., Houghton, P. J., and Wooten, R. S. Mechanism ofinduction of the gastrointestinal toxicity in the mouse by 5-fluorouracil,

5-fluorouridine, and 5-fluoro-2’-deoxyuridine. Cancer Res., 39: 2406-

2413, 1979.

7. Keyomarsi, K., and Moran, R. G. Folinic acid augmentation of theeffects of fluoropyrimidines on murine and human leukemia cells.

Cancer Res., 46: 5229-5235, 1986.

8. Matthews, D. A., Appelt, K., Oatley, S. J., and Xuong, N. H. Crystal

structure of Escherichia coli thymidylate synthase containing bound

5-fluoro-2’-deoxyuridylate and lO-propargyl-5,8-dideazafolate. J. Mol.Biol., 214: 923-936, 1990.

9. Varney, M. D., Marzoni, G. P., Palmer, C. L., Deal, J. G., Webber,S., Welsh, K. M., Baquet, R. J., Bartlett, C. A., Morse, C. A., Booth,C. L., Herrmann, S. M., Howland, E. F., Ward, R. W., and White, J.

6 F. M. Muggia, R. Koda, S. Jeffers, C. G. Leichman, M. Rogers, L.

Leichman, A. Johnston, and N. Clendeninn. Phase I and pharmacologic

study of AG 33 1 : the thymidylate synthase inhibitor on single and daily

x5 schedules, submitted for publication.



1692 Phase I Trial of AG33 1

Crystal-structure-based design and synthesis of benz[c,d]indole-contain-ing inhibitors of thymidylate synthase. J. Med. Chem., 35: 663-676,1992.

10. Webber, S., Shetty, B., Johnston, A., Welsh, K., Varney, M., Deal,

J., Morse, C., and Soda, K. In vitro properties and antitumor activity of

AG331, a novel lipophilic thymidylate synthase inhibitor. Proc. Am.Assoc. Cancer Res., 33: 413, 1992.

11. Shetty, B., Kosa, M., Soda, K., and Webber, S. Pharmacokineticsand tissue distribution of two novel thymidybate synthase inhibitors,AG331 and AG-337 in the rat. Proc. Am. Assoc. Cancer Res., 34: 394,

1993.

12. Miller, A. B., Hoogstraten, B., Staquet, M., and Winkler, A. Re-porting results of cancer treatment. Cancer (Phila.), 47: 207-214, 1981.

13. Qian, M., O’Dwyer, P. J., Clendeninn, N., and Gallo, J. M. High-performance liquid chromatographic method for the determination of

AG33 1 , a novel anticancer agent, in human plasma and urine using solidphase extraction and photo diode-array detection. J. Chromatogr., 666:307-314, 1995.

14. Rocci, M. L., Jr., and Jusko, W. J. LAGRAN program for area andmoments in pharmacokinetic analysis. Comput. Programs Biomed., 16:

203-216, 1983.

15. Gibaldi, M., and Perrier, D. Pharmacokinetics, Ed. 2. New York:Marcel Dekker, 1982.

16. Perrier, D., and Mayersohn, M. Noncompartmental determinationof the steady-state volume of distribution for any mode of administra-tion. J. Pharm. Sci., 71: 372-373, 1982.

17. Lam, F. C., Hung, C. T., and Perrier, D. Estimation of variance forharmonic mean half-lives. J. Pharm. Sci., 74: 229-231, 1985.

18. Rafi, I., Taylor, G. A., Calvete, J. A., Boddy, A. V., Balmanno, K.,Bailey, N., Lind, M., Calvert, A. H., Webber, S., Jackson, R. C.,Clendeninn, N., and Newell, D. R. Clinical pharmacokinetic and phar-macodynamic studies with the non-classical antifolate thymidylate syn-

thase inhibitor, 3,4-dihydro-2-amino-6-methyb-4-oxo-5-(4-pyridythio)-quinazolone dihydrochloride (AG-337) given by 24 h continuousinfusion. Clin. Cancer Res., 1: 1275-1284, 1995.

19. Larsen, R. J., and Marx, M. L. An introduction to mathematicalstatistics and its applications, Ed. 2. Englewood Cliffs, NJ: Prentice-Hall, 1986.

20. Giantonio, B., Qian, M., Gallo, J. M., DeMaria, D., Legerton, K.,Johnston, A. L., Clendeninn, N. J., and O’Dwyer, P. J. Phase I trial ofAG331 as a 5-day continuous infusion. Proc. Am. Soc. Clin. Oncol., 14:

480, 1995.

21. Jackson, R. C., Jackman, A. L., and Calvert, A. H. Biochemicaleffects of a quinazoline inhibitor of thymidylate synthase, N-(4-(N-((2-

amino-4-hydroxy-6-quinazolinyl)methyl)prop-2-ynylamino)benzoyl)-L-glutamic acid (CB3717) on human lymphoblastoid cells. Biochem.Pharmacol., 32: 3783-3790, 1983.

22. Mitrovski, B., Pressacco, J., Mandelbaum, S., and Erhlichman, C.Biochemical effects of folate-based inhibitors of thymidylate synthase in

MGH-Ul cells. Cancer Chemother. Pharmacol., 35: 109-1 14, 1994.

23. Curtis, N. J., Harris, A. L., and Aherne, G. W. Mechanisms of celldeath following thymidylate synthase inhibition: 2’-deoxyuridinetriphosphate accumulation, DNA damage, and growth inhibition follow-ing exposure to CB3717 and dipyridamole. Cancer Res., 51: 2346-

2352, 1991.

24. Jackman, A. L., Taylor, G. A., Calvert, A. H., and Harrap, K. R.Modulation of antimetabolite effects: effects of thymidine on the effi-cacy of the quinazoline-based thymidylate synthase inhibitor, CB3717.Biochem. Pharmacol., 33: 3269-3275, 1984.

25. Heidelberger, C., Danenberg, P. V., and Moran, R. G. Fluorinatedpyrimidines and their nucleotides. In: A. Meister (ed), Advances inEnzymology and Related Areas in Molecular Biology, pp. 57-119. NewYork: John Wiley and Sons, 1983.

26. Berger, S. H., and Hakala, M. T. Relationship of dUMP and freeFdUMP pools to inhibition of thymidylate synthase by 5-fluorouracil.Mol. Pharmacol., 25: 303-309, 1984.

27. Advanced Colorectal Cancer Metaanalysis Project. Modulation offluorouracil by leucovorin in patients with advanced colorectal cancer:evidence in terms of response rate. J. Clin. Oncol., 10: 896-903, 1992.

28. Jones, T. R., Calvert, A. H., Jackman, A. L., Brown, S. J., Jones, M.,

and Harrap, K. R. A potent antitumor quinazoline inhibitor of thymi-dylate synthase: synthesis, biological properties and therapeutic resultsin mice. Eur. J. Cancer, 17: 1 1-19, 1981.

29. Calvert, A. H., Alison, D. L., Harland, S. J., Robinson, B. A.,Jackman, A. L., Jones, T. R., Newell, D. R., Siddik, Z. H., Wiltshaw, E.,McElwain, T. J., Smith, I. E., and Harrap, K. R. A Phase I evaluation ofthe quinazoline antifolate thymidylate synthase inhibitor N’#{176}-propargyl-5,8-dideazafolic acid, CB 3717. J. Clin. Oncol., 4: 1245-1252, 1986.

30. Rafi, I., Taylor, 0. A., Calvete, J. A., Balmanno, K., Boddy, A. V.,Bailey, N. B., Lind, M. J., Newell, D., and Calvert, A. H. A Phase Iclinical study of the novel antifolate AG-337 by a 5-day continuous

infusion. Proc. Am. Assoc. Cancer Res., 36: 240, 1995.

31. Clarke, S. J., Ward, J., DeBoer, M., Planting, A., Verweij, J.,

Sutcliffe, F., Azab, M., and Judson, I. R. Phase I study of the new

thymidylate synthase inhibitor Tomudex (ZD1694) in patients with

advanced malignancy. Ann. Oncol., 5 (Suppl. 5): 132, 1994.

32. Grindey, 0. B., Shih, C., Barnett, C. J., Pearce, H. L., Engelhardt,J. A., Todd, G. C., Rinzel, S. M., Worzalla, J. F., Gossett, L. S., Everson,T. P., Wilson, T. M., Kobierski, M. E., Winter, M. A., Kuhn, D., Taylor,E. C., and Moran, R. 0. LY23l5l4, a novel pyrrolopyrimidine antifolatethat inhibits thymidylate synthase. Proc. Am. Assoc. Cancer Res., 33:

441, 1992.



1996;2:1685-1692. Clin Cancer Res P J O'Dwyer, P B Laub, D DeMaria, et al. 5-day continuous infusion.Phase I trial of the thymidylate synthase inhibitor AG331 as a

Updated version

http://clincancerres.aacrjournals.org/content/2/10/1685

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/2/10/1685To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Phase I Trial of the Thymidylate Synthase Inhibitor …clincancerres.aacrjournals.org/content/clincanres/2/10/...Vol. 2, 1685-1692, October 1996 Clinical Cancer Research 1685 Phase