[CANCER RESEARCH 55, 862-867, February 15, 1995]

formed by hydrocyanation of quinocarmycin (8, 9), were evaluatedin this human tumor cell line screen to identify potential diseasespecific cytotoxicity. Both agents were amongst the first identifiedby the screen to demonstrate melanoma specificity. It is the purpose of the present communication to report the findings of tumorspecificity in vitro and the results of follow-up studies conductedin staged s.c.-implanted human melanoma xenograft modelsderived from the cell lines. A preliminary account of these studieshas been published (10).

MATERIALS AND METHODS

KW2152 (NSC 601422; Mr523) and DX-52-1 (NSC 607097; Mr 357) wereprovided to the N@ by Kyowa Hakko Kogyo Co. Ltd., Tokyo, Japan. Struclures are shown in Fig. 1. Synonyms for KW2152 at earlier stages of devel

opment were quinocarcin, DC-52, and quinocarmycin. For cell culture studies,both agents were solubiized in DMSO at 400 times the desired final maximum

test concentration, and aliquots were frozen at —70°C.Just prior to use,aliquots were thawed at room temperature for 5 mm and then diluted with cellculturemediumcontaining50 @g/mlgentamicinto twice the desiredfinalconcentration. For administration to mice, KW2152 was dissolved in water forinjection, and DX-52-1 was prepared in a 0.9% NaCI solution. The latterpreparations were either smooth suspensions, cloudy, or clear, depending upondosage level.

Tumors and Mice. The human tumor cell lines used in the in vitro screen

and their origins, characterization, and maintenance have been presentedpreviously (11, 12). All cell lines were grown in RPM! 1640 supplementedwith 5% heat-inactivated fetal bovine serum and 2 mM glutamine. For establishment of tumors in mice, the human tumor lines were CUltUredin RPMI1640 supplemented with 10% heat-inactivated serum. Harvested ceHs wereimplanted s.c. in athymic NCr nu/nu mice obtained from the NU animalprogram (Frederick, MD), and the reSulting solid tumors were maintained by

serial in vivo passage of tumor fragments. Mice were housed in sterile polycarbonate, filter-capped Microisolator cages (Lab Products, Inc.), maintainedin a barrier facility on 12-h light/dark cycles, and provided with sterilized foodand water ad libitum.

In Vitro Evaluation. Assays were conducted according to published procedures (12). Briefly, cells were inoculated in 96-well culture plates and

incubated for 24 h prior to the addition of KW2152 or DX-52-1 at five 10-folddilutions. Final test agent concentrations ranged from 0.01 to 100 @.&M.Following a 48-h incubation, cell cultures were fixed in situ with trichloroaceticacid, washed, and then stained with sulforhodamine B. The solubilized stainwas measured spectrophotometrically at 515 nm.

Analysis of differential cytotoxicity was conducted according to published

methods(13).Thelog10molardrugconcentrationthatcauseda 50%reductionin the measured protein at the end of the drug incubation compared with thatat the beginning (LCso)was determined for each cell line. Differences betweenthe log10 @5Oconcentration for each cell line and the mean value for all celllines for the same test agent were plotted (mean graph).

The COMPARE Algorithm. The pattern-recognitionalgorithmtermedCOMPARE (13) has become an extremely valuable analytical tool for identifying agents with similar mechanisms of action based on biological dataobtained from the NC! human tumor cell line screen (14, 15). It is designed tofind mean graph pattern matches to compounds of interest. The compound ofinterest is selected as a seed for a probe of the database. The program thenranks the similarity of the patterns of differential cytotoxicity (mean graph) ofeither all other compounds in the database or a group of approximately 175

862

Efficacy of the Quinocarmycins KW2152 and DX-52-1 against Human MelanomaLines Growing in Culture and in Mice'

Jacqueline Plowman,2 Donald J. Dykes, Ven L. Narayanan, Betty J. Abbott, Hiromitsu Saito, Tadashi Hirata,and Michael R. Grever

Developmental Therapeutics Program, Division of Cancer Treatment. National Cancer Institute. Bethesda, Maryland 20892 Ii. P., V. L N., B. J. A., M. R. G.J: SouthernResearch institute, Birmingham, Alabama 35255 ID. J. Dl; and Kyowa Hakko Kogyo Co., Ltd., Pharmaceutical Research Laboratories, Shizuoka, Japan (H. S., T. H.]

ABSTRACT

Quinocarmycin monocitrate (KW2152) and its analogue, DX-52.1, deanonstrated Specificityfor melanomas in the National Cancer InstitUte in vibvhuman tumor cell line drug screen. In contrast to most cell lines, a 50%reduction in tumor cell burden (as measured protein) at the end of a 48-hdrug incUbatiOnwas produced in five of eight melanoma lines by KW2152concentrations (LC@s) ranging from 0.49 to 10.93 pi@tand by DX-52.1concenfradoan ranging from 0.71 to 7.33 pM. Using the COMPARE algorithm, the patterns ofdifferential cytotoxicityfor both agents at the LC@levelof effect most closely resembled those for actinomycin D, mithramydn, andAdriamydn. In in vivo studies, both KW2152 (40 mglkg/day) and DX-52-1(90 mg/kg/day) mused partial and complete regressions of staged s.c.-im

@tedLOX IMVI melanoma xenografts following @p.administration ondays 5, 9, and 13 and produced tumor growth delays of 231 and 181%,respectively (P < 0001). Activity was augmented by more prolonged therapy.Statistically significant growth inhibitiOn of SK-MEL-2, UACC-62, UACC257,andM14,but notSK-MEL-5andMALME-3M,melanomaxenograltsalso was observed following every fourth or seventh day Lp treatments.Basedon thesefindings,DX-52.1has beenselectedby the NationalCancerInstitute for development to clinical trial especially against melanomas. Thisagent representsone of the first to be selected for preclinicaldevelopmentbased on disease-@ specificitydiscovered in the National Cancer InStitUtecancer drng screen.

INTRODUCTION

A novel antibiotic, quinocarmycin, demonstrating moderate activityagainst Gram-positive but not Gram-negative bacteria, was isolatedfrom culture broths of Streptomyces melanovinaceus (1, 2). Preliminary data indicated that quinocarmycin also inhibited growth of themurine P388 leukemia in mice (1). Further in vivo antitumor evaluation using quinocarmycin monocitrate (KW2152) showed thatKW2152 demonstrated little or no activity against five other murinetumors (3) but was effective against a variety of human tumor xcnografts originating from breast, colon, stomach, and lung (3, 4).KW2152 also was effective against human tumor cell lines in vitro,significantly inhibiting colony formation of three non-small cell lungcancer lines (5) and inhibiting the growth of several human leukemiasand lymphomas (6). Based on its demonstrated preclinical antitumoractivities, clinical studies were initiated with KW2152 (4).

In 1985, the NCI3 initiated a new screening strategy for discoveringnew antitumor agents (7). With the implementation of a 60 humantumor cell line screen involving lines derived from seven cancertypes (brain, colon, leukemia, lung, melanoma, ovarian, and renal),the emphasis shifted from drug-oriented to disease-oriented drugdiscovery. Both KW2152 and DX-52-1, a more stable analogue

Received 9/1/94; accepted 12/16/94.The costs of publicationof this articlewere defrayedin partby the paymentof page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1 Some of these studies were supported by Contract NOi-CM-97553 from the DevelopmentalTherapeuticsProgram,Divisionof CancerTreatment,NationalCancerInstitute.

2 To whom requests for reprints should be addressed, at NCI-FCRDC, Fairview

Center,Suite205,1003W. 7thStreet,Frederick,MD21701-8527.3 The abbreviations used are: NCI, National Cancer Institute; CR, complete tumor

regression; PR, partial tumor regression.

Research. on November 26, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

EFFICACY OF QUINOCARMYCINSAGAINST HUMAN MELANOMA LINES

terms of PRs, CRs, and tumor-free mice. The importance of using tumorregression in animal models as an end point of clinical relevance has beenpropounded by several investigators (17—19).Regressions were defined aspartial if the tumor weight decreased to 50% or less of the tumor weight at

start of treatment without dropping below 63 mg. Both CRs and tumor-freesurvivors were defined by instances in which tumor burden fell below

measurable levels (<63 mg) during the experimental period. The twoparameters differed by observation either of tumor regrowth (CR) or noregrowth (tumor-free) prior to the final observation day. Statistical analysesof both the A tumor weights and the times for tumors to increase by twodoublings (four for the LOX IMVI experiments) were conducted using theMann Whitney rank-sum test.

Toxicitywas assessedon the basisofapparent drug-relateddeathsand net bodyweight loss. Death in a treated animalwas presumedto be related to treatmentifthe animal died within 15 days of last treatmentand either its tumor weight wasless than the lethal tumor burden in the control mice or its net body loss at deathwas 20% greater than the mean net weight change of the controls at death orsacrifice. In each group, a net body weight loss on any observation day wascalculatedas a percentagerelativeto the mean net body weight on the day of firsttreatment.Maximumweight loss and the day on which it occurredwere recorded.Antitumoreffectsare reportedfor treatmentsproducingoptimalactivity,providedtherapydid not cause more than one death per group.

RESULTS

Differential Cytotoxicity. KW2152 and DX-52-1 were evaluatedin the NC! human tumor cell line screen. Fig. 2 presents a visualimage of differential cytotoxicity (mean graph) obtained from thescreening results based on log10 molar concentrations of the quinocarmycins that caused a 50% reduction in protein in each cell line.Each mean graph represents average results from two experiments.Cell lines more sensitive than average are depicted by bars projectingto the righr, cell lines less sensitive than the mean are shown by barsprojecting to the left. Specificity for melanoma lines was observed forboth agents with six or seven of eight melanomas demonstratinggreater sensitivity than the average cell line. Specificity for melanomas was not absolute; one or two sensitive tumor lines were noted inother histological groupings. For most cell lines, a 50% reduction inprotein was not attained at the highest concentration evaluated (100p.M; depicted by truncated bars on the left of the mean graph). When

the patterns of differential cytotoxicity were compared to those for175 standard agents, the agents providing the ten closest matches(Table 1) were predominantly DNA-binding agents, alkylating agents,

and two DNA binders with alkylating potential. Correlation coefficients were all greater than 0.62. Although the patterns for KW2152and DX-52-1 were not identical, three of four of the closest matchesfor both agents were the same DNA-binding agents: actinomycin D,mithramycin, and Adriamycin. When KW2152 and DX-52-1 wereused to probe the much larger database of more than 8000 compounds,the closest match for each agent was a DNA binder, aclacinomycinhydrochloride for KW2152 and nogalamycin for DX-52-1 with correlation coefficients of 0.882 and 0.887, respectively (data notshown). Other close matches included three saframycin analogues,with correlation coefficients ranging from 0.835 to 0.864 for KW2152and from 0.737 to 0.864 for DX-52-1.

Antitumor ACtiVity against Melanoma Xenografts. The antitumor efficacies of i.p.-administered KW2152 and DX-52-1 were eva!-uated in seven s.c.-implanted human melanoma xenogra.ft models, sixof which were derived from the cell lines used in the in vitro screen.Growth characteristics of tumors in the vehicle-treated control micefrom the efficacy experiments are summarized in Table 2. Four of theseven tumors had initial doubling times of approximately five days.The LOX IMVI amelanotic melanoma grew more rapidly, with adoubling time around 1.5 days, and the M14 and MALME-3M melanomas grew more slowly, with doubling times of 7.05 and 11.7 days,

H

%TIC=@ x 100

%TIC= @-x 100

where @T<0 and T1is the median tumor weight at the start of treatment. Theoptimum (minimum) value obtained after the end of treatment is presented.Tumor growth delay was expressed as a percentage by which treated groupweight was delayed in attaining a specified number of doublings compared tocontrols using the formula [(T-C)ICJ x 100, where T and C are the mediantimes in days for treated and control tumors, respectively, to attain thespecified size (excluding tumor-free survivors and drug-related deaths). Areduction of tumor burden during the experimental period was recorded in

863

?H2CO2H

HOCCO2H

CH2OH

KW2152

DX-52-1

Fig.1. Structuresof quinocarmycinmonocitrate(KW2152)and its hydrocyanationproduct,DX-52-1.

standard agents composed of approved anticancer chemotherapeutic agents,Investigational New Drugs, and those under development by the NC! to thepattern of the seed compound. The analysis uses a SAS statistical program forcalculating Pearson correlation coefficients; a correlation coefficient of 1.0signifies identical patterns.

In Vii'oEvaluations. Tumorfragments(about30 mg each)were implantedS.C. into the axillary region of athymic mice on day 0. Treatment was initiated

when tumorsreachedspecified weight rangesindicatedin the tables. Tumorweights were calculated from in situ caliper measurements of tumor length andwidth using the formula for a prolate ellipsoid (16) and assuming that thespecific gravity of the tumor was not significantly different from 1. Tumor sizeand body weights were recordedapproximatelytwo times per week. BothKW2152 and DX-52—1 were administered i.p. at multiple dose levels onintermittent schedules, either every fourth or seventh day for three or sixtreatments. All treatments were administered by exact body weight (0.1 mi/lOg) to groupsof 6 or 10 mice; 20 vehicle-treatedtumorouscontrolmice wereincluded in each experiment.

Three parameters were used to evaluate antitumor effects: tumor inhibition,tumor growth delay, and tumor regression as denoted by % TIC, % (.T-C)/C,and PR and CR, respectively. Changes in tumor weight (@ weights) for each

treated (7) and control (C) group were calculated for each day tumors weremeasured by subtracting the median tumor weight on the day of first treatmentfrom the median tumor weight on the specified observation day. These valueswere used to calculate a percentage T/C as follows:

where i@T > 0, or

Research. on November 26, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

L•uk@miaCCRF-CEMHL-60(TB)K-562MOLT- 4RPMI-8226Non—Small Cell LungA549/ATCCEKVX —HOP-18 —HOP-62 —HOP-92 —NCI-H226 - UNCI-H23 —NCI-H322MNCI-H460 — —NCI-H522 —Small Cell Lung Canc.r_____________________DMS 114 —DMS 273 —Colon Cancer ___________________ _ _COLO 205 —DLD-1 —HCC-2 998HCT-116 -HCT-15 —HT29 —KM12 —KM2OL2 -SW-620CNS Canc•r_____________________________SF-268 — —SF-295 — —SF-539 — —SNB-19 — —SNB-75 — —SNB-78 — —U251 —XF 498 — —MelanomaLOX IMVIMALME-3MM19 -MELSK-MEL-2SK-MEL-28SK-MEL-5UACC-257UACC- 62Ovarian Cancer_ _ _IGROV1OVCAR- 3OVCAR-4OVCAR-5OVCAR-8 —SK-OV-3 —R•nal Canc•r___________________________@@A498 —CAKI-1 —RXF-393SN12C — —UO-31 - —

I—

EFFICACY OF QUINOCARMYCINS AGAINST HUMAN MELANOMA LINES

DX—52-1

—T

Fig. 2. Patterns of differential cytotoxicity towardhumantumorcell lines.Thecytotoxiceffectsof the two quinocarmycinswereevaluatedagainst53 human tumor cell lines comprising the NCIdrog-discoveryscreen.Foreachcell line, the log,0molar concentration producing 50% cytotoxicity(LC50) was obtained by extrapolation of datapoints. A mean of the individual log,0 LC@s wascalculatedfor eachagentand usedas the anchorpoint (center line) for this graphic presentation(meangraph).Then the individualresponseof eachcell line to either KW2152 or DX-52-1 was depictedby a bar extendingeither to the right or leftofthe mean,dependingonwhetherthecell linewasmoreor lesssensitivethan the averageresponse.The datarepresentthe averageof two experiments.The lengthof eachbar is proportionalto the relative sensitivity compared with the mean determination. Truncated bars to the left indicate that 5096cytotoxicitywas not attainedat the highestconcentration evaluated(100 pM). A dashedline (--) onthe axis indicatesthat the cell line wasnot successfully testedwith the agent.

Differential Cytotoxicity(Relative Deviation from Mean)

respectively.As a resultoftumor growth andhandling,mostcontrol micelost 5-10% body weight during the experimental period. Weight loss in

mice bearing UACC-257 melanomaswas more marked (22.7%).Both quinocarmycins demonstratedantitumor activity in five of the

sevenmodels.The greatestefficacy wasobservedagainstthe rapidlygrowing LOX IMVI melanomas. Following i.p. administration ofeither 40 mg/kg KW2152 or 90 mg/kg DX-52-1 on days 5, 9, and 13,tumors regressed until a nadir was reached on day 15, after whichregrowth occurred at a rate similar to the original growth rate ob

served in vehicle-treated control mice (Fig. 3A). KW2152 caused 1 of10 CRs by day 19, with a durationof 16.8 days,and6 of 10 PRs,witha mediandurationof 5.7 days.DX-52-1 caused2 of 10 CRs and3 of10 PRs, with median durations of 5.9 and 6.3 days, respectively.

Tumor growth delaysbasedon time to four doublingswere 231 and181% for KW2152 and DX-52-1, respectively (P < 0.001 for both,

Mann Whitney rank-sum test). Growth delays of 74, 123, and 169%wereattainedfollowing treatmentwith dosesof 11.9,17.8,and26.7mg KW2l52Ikg/day, respectively,and104, 109, and 174% with doses

864

Human TumorCell Lines KW2152

Research. on November 26, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

Table 1 Ranking ofdifferential cytotoxicitypatterns usingeither KW2152 orDX-52-1asprobes in the COMPARE pattern-recognitionprogramBoth

KW2152andDX-52-1 wereusedasseedsto probethestandardagentdatabasecomprisedof approximately 175 clinical drugsapprovedfor the treatmentof cancerandNCI

investigationaland developmentalantitumor agents.The COMPAREalgorithmrankedall compounds in this database according to the similarity of their meangraphpatternsto the meangraphpattemof eachquinocarmycin.The patternscomparedwerethose

obtainedat the ‘@50level of effect as illustratedfor KW2152 and DX-52-1 inFig.2.The ten compoundswith patternsmost closely matchingeach quinocarmycinarepresented.Seed:

KW2152 Seed:DX-52-1Correlation

CorrelationClosestmatches coefficient ClosestmatchescoefficientActinomycin

D 0.877 Mithramycin0.788Mithramycin0.785 Actinomycin D0.776Uracil

N2 Mustard 0.771 Adriamycin0.736Adriamycin(ADR) 0.769 Phyllanthoside0.720Cyanomorpholino-ADR

0.741 Morpholino-ADR0.699ChromomycinA3 0.736 Cyanomorpholino-ADR0.696Daunomycin

0.714 Daunomycin0.694Chlorambucil0.704 ChromomycinA30.687Morpholino-ADR0.700 Echinomycin0.648Mitomycin

C 0.695 AclacinomycinA 0.629

Onetumorfragment(approximately30mg)wasimplanteds.c.into theaxillaryregionofeachathymicNCr nulnu mouseon day 0. Tumor growth was monitored.Micewererandomized,

and treatmentswith 0.9% NaCl (vehicle control) were initiated oneveryfourthor seventhday schedules,when tumorsattainedthe weight rangeshowninthetable.

All control tumorsgrew after the mice were randomized(20/experiment), withtheexceptionof oneeachin the M14 andMALME-3Mexperiments.Time

in Max. %rd.Doublingdays to Tumor wt. meannettime

in two rangeat Staging wt. lossMelanoma days0 doublings― staging(mgf day(day)―LOX

IMVI 1.4 2.9 100—405 5 14.2(15)1.753.7 100—365 59.0(15)SK-MEL-2

5.1 10.6 100—405 144.2(21)SK-MEL-55.6 13.1 100—405 15 7.8(47)UACC-625.1 11.0 100—392 167.8(34)UACC-2574.9 16.1 88—352 1622.7(41)M147.05 24.1 100—234 12 11.1(22)MALME-3M

11.7 46.5 108—405 27 5.6(51)

EFFICACY OF QUINOCARMYCINS AGAINST HUMAN MELANOMA LINES

of 26.7, 40, and 60 mg DX-52-lIkg/day, respectively (data notshown). In a second study, more prolonged treatment(every fourthday for six treatments) augmented the antitumor effects of bothquinocarmycins (Fig. 3B). The 40 mg/kg/day dose of KW2152 was

equal to, or greater than, a maximally tolerated dose, causing one ofsix drug-related deaths on day 12 and a 24% mean net body weight

loss on the same day. In the remaining five mice, three CRs wereattained with a median duration of 27.7 days, and a fourth mouse wastumor-freefrom day 12 throughthe end of the study on day 71. Thetumorgrowthdelay was 403% (P < 0.001). The 26.7 mg/kg/daydosecaused no deaths and less body weight loss but still produced one ofsix PRs, oneof six CRs, andtwo of six tumor-freemice, anda tumorgrowth delay of 313% (P < 0.001; data not shown). Growth delays of116 and 221% (latter responseincludes two CRs) were attainedfollowing treatment with 11.9 and 17.8 mg/kg/day, respectively. The60 mg/kg/day dose of DX-52-1 produced four of six CRs, with amedian duration of 20.8 days and a growth delay of 389%

(P = 0.002). Some host toxicity was associated with the lattertreatment: one of six apparent drug-related deaths on day 29 and a

10000A

ft IP Rx

10 15 20 25 30

t

@L I I

0 5

B

Table 2 Growth characteristicsof s.c-implanted human melanomaxenograft controlsusedin the efficacystudies

Fig. 3. Response of advanced stage s.c-implanted human LOX-IMVI melanomaxenografts to i.p. KW2152 and DX-52-1. About 30 mg tumor fragments were implanteds.c into theaxillary regionof athymicNCr nulnuathymicmiceon day0. Treatmentwasinitiated on day 5. A, Q4Dx3 schedule:U, controls treatedwith 0.9% NaCI; •,90mg/kg/dayDX-52-1;A, 40 mg/kg/dayKW2152.B. Q4Dx6schedule:U, controlstreatedwith 0.9%NaCl; •,60 mg/kg/dayDX-52-1; A, 40 mg/kg/dayKW2152.

maximum net body weight loss of 15.5% recorded on day 8. Lowerdosesof 26.7 and 40 mg/kg/day both produced one of six tumor-freemice and growth delays of 104 and 175%, respectively.

Statistically significant antitumor effects were observed in four ofthe remaining six melanoma xenograft models evaluated (Table 3).Activity was similar for both agents. Complete inhibition of tumorgrowth was observedon day 32 after administrationof 40 mgKW2152/kg on days 14, 21, and 28 to mice bearing SK-MEL-2 tumors.This antitumor effect was associatedwith marked net body weight loss.However, the 26.7 mg/kg/day dose also causedtumor stasiswithoutproducing body weight loss in excessof that observedin controls (Table2). Two CRs were attained during treatment with a median duration of6.53 days,and a good tumor growth delay of 185% was measured.TwoCRs of approximately 6.18 days duration also were attained following

a Median time for control tumors to grow from 200 to 400 mg.

b Median time for control tumors to grow to four times their staging day weight.C Tumor weight range for all mice used in the experiment on the first day of treat

ment (staging day).d Maximum mean net body weight loss in control mice as a percentage of the staging

day weight. The number in parenthesesindicates the day on which the maximum weightlosswasrecorded.No calculationswereconductedwhenmorethan33%of themicewerelost to evaluationdueto tumor growth andresultingsacrifice.

865

a)

)@DaysAfter Implantation

10

10000

a)

@ 1000

.@2100-a

10

. Of Measurability

f t f t t flPRx

0 10 20 30 40 50DaysAfterImplantation

Research. on November 26, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

Table 3 Responseof stageds.c-implanted humanmelanomaxenograftsto i.p.-administeredKW2152 andDX-52-1One

tumor fragment(about30 mg) was implanteds.c. into theaxillary regionof eachathymicNCr nu/numiceon day 0. Treatmentwas initiatedwhen tumorsattainedsizesrangingfromapproximately 100 to 400 mg (see Table 2). Each agent was evaluated at four dosage levels ranging from 11.9 to 40 mg/kg/day for KW2152 and from 26.7 to 90 mg/kg/dayforDX-52-1.

Optimal antitumoreffectsare presented.Dose

Max. 96 rd. mean net wt. Optimal AT/AC Growth delay:Melanoma Treatmentdays Testagent (mg/kg/day) loss(day) (day)a pb % (T —C)/CSK-MEL-2

14,21,28 KW2152 40 26.8(18) 0 (32) <0.001 214<0.00126.75.0(25) 3(32) <0.001 185<0.001DX-52-1

90 7.5(32) 10(32) <0.001 118<0.001SK-MEL-515,19,23 KW2152 40 10.2(26) 61(33) NSC 51NSDX-52-1

60 21.1(26) 54(33) <0.05 56NS4012.7 (26) 49(33) NSC 27NSUACC-62

16,23,30 KW2152 26.7 13.0(51) 11 (34) <0.05 112<0.05DX-52-190 9.1 (48) 18 (34) <0.005 185<0.05UACC-257

16,20,24 KW2152 40 13.5(27) 5 (27) <0.001 95<0.001DX-52-190 17.5(27) 12(27) <0.005 35<0.005M14

12,16,20 KW2152 26.7 16.0(47) 7 (22) <0.01 —2NSDX-52-190 16.5 (26) 19 (26) <0.05 56NSMALME-3M

27,31,35 KW2152 26.7 8.7(41) 48(72) NSC —2NSDX-52-190 13.6(41) 64(72) NSC 4NSa

MinimumpercentageTICbasedonchangesintumorweightattainedaftertreatmenthadended.Thenumberinparenthesesindicatesthedaypostimplantonwhichtheeffectwas

EFFICACY OF QUINOCARMYCINS AGAINST HUMAN MELANOMA LINES

attained.b Mann Whitney rank-sum test.C NS, not significant; P > 0.05.

treatment with the 90 mg/kg/day dose of DX-52-1, and the mediangrowth delay was 118%. The response of s.c.-implanted UACC-62melanomaswas similar to that of the SK-MEL-2 tumors, except that noPRs or CRs were attained. At maximally tolerated doses, KW2152 and

DX-52-1 inhibited tumor growth by 89 and 82%, respectively, andproducedgood growth delays of 112 and 185%. Similar growth inhibitions ranging from 81 to 95% were observedin the UACC-257 and M14melanomamodelsbut, asindicatedby the smallermediangrowth delays,the long-term effects of KW2152 against M14 melanomasand of DX52-1 againstboth tumors were not aspronounced.No significant antitumor effects were observedin the SK-MEL-5 and MALME-3M modelsfollowing administration of either quinocarinycin at dosagelevels thatcausedless than 20% net body weight loss.

DISCUSSION

In 1985, the NC! decidedto use humantumor cell lines in an invitro assay as the primary screen for discovering new potential antitumor agents(7). The new screenchangedthe approach for anticancerscreening from compound-oriented to disease-panel oriented drugdiscovery. Compounds demonstrating disease-panelspecificity in thescreen could be subsequently examined further by in vivo modelsderived from the most sensitive in vitro tumor lines. This screen has

been fully operational since April 1990 (20). Two of the first agentsdemonstrating disease-specific activity in the screen were the quinocarmycins, KW2152 and DX-52-1; six (KW2152) or seven (DX52—1)of eight melanoma lines were more sensitive than the averagecell line at the LC50 level of effect (Fig. 2). Although specificity formelanomas was not absolute and probably not to be expected fromprevious studies with KW2152 (5, 6), only one or two tumors in otherhistological groups exhibited similar sensitivity. Furthermore, antitumor efficacy was demonstrated against melanomas in vivo usingstageds.c-implanted tumor models, someof which were derived fromthe same melanoma lines for which sensitivity was observed in vitro.Not unexpectedly considering differences between the two assays,adirect relationship between in vitro cytotoxicity and in vivo activitywas not observed (Table 4). The best in vivo efficacy was observedagainst the rapidly dividing LOX IMVI melanoma xenografts, a!though the tumor line did not appearto be sensitive to KW2152 in thein vitro assay. However, only one parameter of in vitro effect ispresented.Althoughdifferentialeffects were not as pronounced,melanoma specificity also was observed at the total growth inhibitory

level of effect. Using the lattereffect level, the LOX IMVI melanomaline was more sensitive than the average cell line in one of twoexperimentsfor eachagent(data not shown).Changingassayconditions, such as assay duration, may also modify response. Using aclonogenic assay, Fiebig et a!. (21) have confirmed specificity ofDX-52—1for humanmelanomalines and have demonstratedin vivoefficacy againsts.c.-implantsof melanomas.

In the current antitumorefficacy evaluations,both KW2152 andDX-52-1 demonstratedsimilar activity againststageds.c.-implantedmelanoma xenografts. Excellent activity was observed against LOXtumorswith at least50% of the tumorsdemonstratinga substantialreduction in size (PRs + CRs + tumor-free) following i.p. administrationof optimaldosesevery 4 daysfor threeor six treatments.Thislevel of activity is greater than that attained with 11 of 12 standardclinical antitumor drugs (22). Only 1,3-bis(2-chloroethyl)-1-nitrosoureaappearedto be more active; dacarbazinewas lesseffective.Good inhibition of tumor growth also was observed against four of six

additional melanoma xenografts (Table 3). These activities, plus themelanoma-specificcytotoxicityobservedin the NCI screen,providesupport for the evaluation of a quinocarmycin derivative in melanomapatients and led the NC!, in collaboration with Kyowa Hakko Kogyo,to select DX-52-1 for development to clinical trial. Previously,KW2152 had received a limited clinical evaluation in Japan (4) butdid not receive evaluation in melanoma patients. It is no longer in

Table4 Comparisonof theeffectsof thequinocarmycinsagainsthumanmelanomasinvitro and in vivo

KW2152DX-52-1

In vitroInvivoInvitroInvivoMelanomaI@C50concn.

()aactivit% ratingLC@concn.

(@activityratingLOXIMVI>100A,3+UECA,3+SK-MEL-23.18A,

2+0.71A,2+SK-MEL-53.83N50.86NUACC-620.49A,

1+4.50A,1+UACC-257

M1484.85 N'@'A,1+

A,1+7.33 NTA,1+

A,1+MALME-3M43.27N5.21N

a Average concentrations from two experiments.

b @,active; 3+: tumor regression, % TIC range —50to —100, 50% PRs + CRs; 2+:tumorstasis,% TIC range0 to —49or % TIC0-10with PR+ CR rate<50%; 1+: tumorinhibition, % TIC range1 to 40 andno PRs+ CRs;N: inactive,% TIC >40.

C UE, unevaluable due to quality control parameters.

d@ not tested.

866

Research. on November 26, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

EFFICACY OF QUINOCARMYCINSAGAINST HUMAN MELANOMA LINES

clinical trial due to toxicities observed following prolonged dailytreatments for 14 days.4The significance of the latter findings for thetoxicological profile of DX-52-1 in humans is unknown but will befurther explored in preclinical toxicological studies in animals. AlthoughDX-52-1 may be convertedto quinocarmycinin the presenceof acid or silver nitrate (8, 9), there is surprisingly little evidence ofinterconversion or instability of DX-52-1 at physiological pH, and

approximately50% of DX-52-1 appearsunchangedin urinefollowingi.v. administrationto mice.5 Thus, DX-52-1 should be considered aseparateentity from KW2152 and may demonstrate a different toxicological profile. Also, KW2152 was less toxic in patients whenadministered on a less frequent regimen (1 to 3 times weekly).4 Thus,baseduponthe review of the humantoxicity observedwith KW2152and the good preclinical antitumor activity observed for both agentswith intermittenttherapy,an interruptedscheduleof drugadministration appears best to avoid toxicity and enhance the likelihood ofantitumorresponsesin patients.

The mechanisms of action of KW2152 and DX-52-1 have not beenfully elucidated. From the databaseof 175 standard agents, the pattern-recognition program COMPARE identified DNA-binding agentsas the closestmatchesto both quinocarmycins,thereby suggestingDNA interaction as a mechanismby which the compoundsmightexert their effects. In this regard,it is of potential importancefor theclinical development of DX-52-1 that the compound may not beaffected by the phenomenon of multidrug resistance. The multidrugresistant subline of P388 leukemia, developed through exposure toAdriamycin, demonstrated collateral sensitivity to DX-52-1 in vivo.6

When the differential cytotoxicity patterns of the quinocarmycinswere compared to those for the larger databaseof over 8000 compounds,severalsaframycins,includingsaframycinA, were identifiedas close matches with correlation coefficients ranging from 0.737 to0.864. The structural similarity of KW2152 to saframycin suggeststhat it may act through a mechanismproposedfor saframycin A:interaction with DNA through formation of a reactive iminium ion(23) and minor groove binding (24). It has been proposed that themode of actionof KW2152 involvesopeningof the oxazolidineringto give an iminium ion, followed by initial noncovalentbindingin theminor groove and subsequent a!kylation of the 2-amino group ofguanine (9, 25). Similar to saframycin A (26), KW2152 preferentiallyblocks RNA synthesis in a murine leukemia (3), while DNA synthesisis more sensitive in Bacillus subtilis (27).

In summary,KW2152 and DX-52-1 havedemonstratedspecificityfor melanomasin the NC! humantumorcell line screenandexcellentto good activity in five staged, s.c.-implanted human melanomaxenograft models. These findings have resulted in the selection ofDX-52-1 for additional preclinical evaluation, including studiesonschedule dependency, pharmacology, and toxicology. The data fromthese studies will provide the basis for designing Phase I/I! clinicaltrials to assessthe clinical effectiveness of the agent and to test thevalidity of the NCI screen.

REFERENCES

1. Tomita,F. T., Takahashi,K., andShimizu,K-I. DC-52,a novelantitumorantibiotic.I. Taxonomy, fermentationand biological activity. J. Antibiot., 36: 463—467,1983.

2. Takahashi,K., and Tomita, F. DC-52, a novel antitumor antibiotic. II. Isolation,physico-chemical characteristics and structural determination. J. Antibiot., 36:468—470,1983.

4 T. Hirata, unpublished data.

5 J. J. McCormack, personal communication.

6 J@ Plowman, unpublished data.

3. Fujimoto, K., Oka, T., and Morimoto, M. Antitumor activity of a novel antitumorantibiotic, quinocarmycin citrate (KW2152). Cancer Rca., 47: 1516—1522,1987.

4. Inoue, S., Kubota,T., Ohishi, T., Kuzuoka, M., Oka, S., Shimoyama,Y., Kikuyama,S.,Ishibiki,K.,andAbe,0. Antitumoractivityof quinocarmycincitrate(KW-2152)againsthumantumor xenograftsserially transplantedinto nudemice. Keio J. Med.,37: 355—364,1988.

5. Jets, J. R., Saijo, N., Hong, W-S., Sasaki, Y., Takahashi, H., Nakano, H., Nakagawa,K., Sakurai,M., Suemasu,K., andTesada,M. Thecolonyinhibitionof a newtherapeuticagent(KW2152)againsthumanlungcancercelllines.Invest.NewDrugs,5: 155—159,1987.

6. Inaba, S., and Shimoyama, M. Antitumor activity of quinocarmycin (KW2152)againstvariousculturedleukemiaandlymphomacell linesin vitro. CancerRes.,48:6029—6032,1988.

7. Boyd,M. R. Statusof theNCI preclinicalantitumordrugdiscoveryscreen.In: V. T.DeVita, Jr., S. Hellman, and S. A. Rosenberg (eds.), Cancer: Principles and Practiceof Oncology,Updates,Vol. 3, pp 1—12.Philadelphia:Lippincott, 1989.

8. Saito, H., and Hirata, T. Synthetic approachto quinocarcin.TetrahedronLett., 28:4065—4068,1987.

9. Saito,H., Kobayasbi,S.,Uosaki,Y., Sato,A., Fujimoto,K., Miyoshi, K., Ashizawa,T.,Morimoto,M.,andHirata,T. Synthesisandbiologicalevaluationof quinocarcinderivatives. Chem. Pharm. Bull., 38: 1278—1285,1990.

10. Plowman,J., Narayanan,V. L., Abbott, B. J., Inoue, K., Hirata, T., andGrever, M. R.Efficacy of quinocarmycinsKW2152 andDX-52—1againsthumanmelanomasgrowing in culture and in mice. Proc. Am. Assoc. Cancer Res., 34: 368, 1993.

11. Stinson,S. F., Alley, M. C., Kopp,W. C., Fiebig,H-H., Mullendore,L A., Pittman,A. F.,Kenney,S.,Keller,i., andBoyd,M.R.Morphologicalandimmunocytochemical characteristicsof humantumorcell linesfor usein a disease-orientedanticancerdrug screen.AnticancerRes.,12: 1035—1054,1992.

12. Monks,A., Scudiero,D., Skehan,P.,ci a!. Feasibilityof a high-flux anticancerdrugscreenusinga diversepanelof culturedhumantumorcell lines.J. NatI.CancerInst.,83:757—766,1991.

13. Paull,K. D., Shoemaker,R. H., Hodes,L, Monks,A., Scudiero,D. A, Rubinstein,L., Plowman,J., and Boyd, M. R. Display and analysis of patternsof differentialactivity of drugs against human tumor cell lines: development of mean graph andCOMPAREalgorithm.J. Natl. CancerInst.,81: 1088—1092,1989.

14. Paull, K. D., Lin, C. M., Malspeis, L., and Hamel, E. Identification of novelantimitotic agentsacting at the tubulin level by computer-assistedevaluationofdifferentialcytotoxicity data.CancerRes.,52: 3892—3900,1992.

15. Paull,K. D., Hamel,E., andMalspeis,L The predictionof biochemicalmechanismof actionfrom the in vitro antitumorscreenof the NationalCancerInstitute.In: W.0. Foye(ed), CancerChemotherapeuticAgents,in press.Washington,D. C.:American Chemical Society, 1995.

16. Geran, R. I., Greenberg,N. H., MacDonald, M. M., Schumacher,A. M., andAbbott,B. J. Protocolsfor screeningchemicalagentsandnaturalproductsagainstanimaltumorsandotherbiological systems(Ed. 3). CancerChemother.Rep.,Part3, 3(2):51,1972.

17. Martin, D. S., Stolfi, R. L., and Sawyer, R. C. Commentary on “clinicalpredictivity of transplantabletumor systems in the selection of new drugs for solidtumors: rationale for a three-stagestrategy.―Cancer Treat. Rep., 68: 1317—1318,1984.

18. Martin, D. S., Balis, M. E., Fisher,B., ci al. Role of murine tumor models in cancertreatment research. Cancer Rex., 46: 2189—2192,1986.

19. Stolfi, R. L, Stolfi, L M., Sawyer,R. C., and Martin, D. S. Chemotherapeuticevaluationusingclinical criteriain spontaneous,autochthonousmurinebreasttumors.J.Natl.CancerInst.,80:52—55,1988.

20. Grever,M. R., Schepartz,S. A., andChabner,B. A. The NationalCancerInstitute:cancerdrug discovery and developmentprogram. Semin. Oncol., 19: 622—638,1992.

21. Fiebig, H. H., Berger, D. P., Dengler, W. A., Drees, M., Mayo, J., Malspeis, L, andGrever,M. CyanocyclinA andthequinocarmycinanalogNSC607097demonstrateselectivityagainstmelanomaxenograftsin vitro andin vivo.Proc.Am. Assoc.CancerRes.,35: 468, 1994.

22. Dykes, D. J., Abbott, B. J., Mayo, J. 0., Harrison,S. D., Jr., Laster,W. R., Jr.,Simpson-Herren,L., andGriswold,D. P.,Jr.Developmentof humantumorxenograftmodels for in vivo evaluation of new antitumor drugs. In: H. Huber and W. Quei8er(eds.),Contributionsto Oncology,Vol. 42, pp. 1—12.Basel:Karger,1992.

23. Ishiguro, K., Takahashi, IC, Yazawa, K., Saldyama, S., and Arai, T. Binding ofsaframycinA, a heterocyclicquinoneanti-tumorantibiotic,to DNA asrevealedby theuseof theantibiotic labeledwith [‘4CJtyrosineor (‘4C)cyanide.J. Biol. Chem.,256:2162—2167,1981.

24. Lown, I. W., Joshua, A. V., and Lee, J. S. Molecular mechanisms of binding andsingle-strandscissionof deoxyribonucleicacidby theantitumorantibioticssaframycm A and C. Biochemistry,21: 419—428,1982.

25. Hill, 0. C., Wunz, T. P., and Remers, W. A. Computer simulation of the binding ofquinocarcinto DNA. Predictionof modeof action and absoluteconfiguration.J.Comput.-AidedMol.Design,2: 91—106,1988.

26. Ishiguro,K., Sakiyama,S., Takahashi,K., andArai, T. Mode of actionof saframycinA, a novel heterocyclic quinone antibiotic: inhibition of RNA synthesis in vivo and invitro. Biochemistry,17: 2545—2550,1978.

27. Tomita,F.,Takahashi,K., andTamaoki,T. Quinocarcin,anovelantitumorantibiotic.III. Modeof action.J.Antibiot.,37: 1268—1272,1984.

867

Research. on November 26, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

1995;55:862-867. Cancer Res   Jacqueline Plowman, Donald J. Dykes, Ven L. Narayanan, et al.   Human Melanoma Lines Growing in Culture and in MiceEfficacy of the Quinocarmycins KW2152 and DX-52-1 against

  Updated version

  http://cancerres.aacrjournals.org/content/55/4/862

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

  .pubs@aacr.orgDepartment 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://cancerres.aacrjournals.org/content/55/4/862To request permission to re-use all or part of this article, use this link

Research. on November 26, 2020. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from