`
`Serum Prostate-Specific Antigen Decline as a Marker of
`Clinical Outcome in Hormone-Refractory Prostate Cancer
`Patients: Association With "Progression-Free Survival, Pain
`End Points, and Survival
`
`By Eric J. Small, Alex McMillan, Marlc Meyer, Licing Chen, William J. Sliehenmyer, Peter F. Lenehan, and Maria Eisenberger
`
`1-Fgse: Validated end points are lacking for clini-
`cal trials in hormone-refractory prostate cancer {HRPC}.
`Controversy remains regarding the utility of a post-
`treatment decline of prostate-specific antigen [PSA]. The
`purpose of this study was to determine whether post-
`treatment declines in PSA were associated with clinical
`measures at improvement in a randomized phase III
`trial of suramin plus hydrocortisone versus placebo
`plus hydrocortisane.
`Patients and Methods: A total of 460 HRPC patients
`were randomized to receive suramin plus hydrocarti-
`sane [n = 229) or placebo plus hydrocortisone {n =
`231}. All patients had symptomatic, metastatic HRPC
`requiring opioid analgesics. Clinical end points evalu-
`ated inclucled overall survival, obiective progression-
`free survival {OPFSL and time to pain progression {ITPP}.
`An evaluation of overall survival, OPF5, and TIFF as a
`function of a PSA decline of 2 50%, lasting at least 28
`days, was undertaken by using a landmark analysis at 6,.
`
`9, and ‘I2 weeks. A multivariate analysis of the impact of
`PSA decline was performed on these clinical end points.
`gfi: A decline in PEA of 2 50% lasting 2 28 clays
`was significantly associated with a prolonged median
`overall survival, OPFS, and ITPP, both in the entire group
`and the suramin plus hydrocortisone group at all three
`landmarks in both univariate and multivariate analysis.
`Qgnclusion:
`In this prospective, randomized trial of
`suramin plus hyclrocartisone versus placebo plus hydro-
`cartisone, a posttherapy decline in PSA of 2 50%, lasting
`28 days, was associated with prolonged median overall
`survival,
`improved median progression-free survival,
`and median 'lTPP. This analysis suggests that a posttreat-
`rnent decline in-PSA may be a reasonable intermediate
`end point in HRPC trials and calls into question the clinical
`utility of preclinical assaysuevaluating the in vitro effect at
`given agents on PSA secretion.
`J Clin Oncol
`I9: l304- I3 I I. 0 2001 by American
`Society of Clinical Oncology.
`
`HE LACK OF v-.tliv:ltItet'l end points for clinical trittlrs in
`hnrnmne-re|'mct0i'y prostate L‘Eil1L‘C1' tl-IRPC) has long
`elmlleiigcd t.‘ll!1lL.‘t‘tl it1xresligtttoi's.' Althmtgh imprcivement in
`HtIl‘VlVi'tl retnuins the gold SlElI‘lt'l€tI‘(l
`tn tlemonstmte CllI'll(_‘[ll
`
`benefit. this end point is not 1133.‘-i{:.‘iSill’Jlt:‘ in phztse ll Httidics
`and m:1_v he difficull to ev:1|u£t1e in |_'tl1£I!iC lll clinical trittl.~;,
`which are l’reqtientl_v cemlotintlctl by L‘I‘(‘t_‘i.‘i~0\-’t'il'
`tlcsigma
`ll.‘-it3Cl to en.~‘.tIre that Elll patients have access in patetitiztlly
`henelicittl tliempy. The lttek of hicliiiteiisiemilly i11e:tstti't1ble
`disettse in more Eli:-iii 7(Il‘/r.- of HRPC patieiits. also precludes
`ttsiilg objective l't‘_‘-1-}|)t'lllt:ltEFi
`in inettsttmble disease its an end
`point.
`POSl'll1€I‘il13)’
`decline of pi‘nst:ite—5apccilie antigen
`{PSAi has been E\";1lLlilit:(l in multiple reports and ltilti been
`rceomtnencled as :1 potential market‘ 0]‘ 1'0.-;|)oI1se.3"' How-
`ever. the lI!~‘t.‘. of :1 decline in PSA as an int'ermu.:t'|i:tte m:1t‘ke1'
`
`lies not been pimipeetively vulitlztted. anti
`of t'e.~;pnnse
`e0nt'roversy l‘l:‘]111ll|1H us to its tttility.'”‘
`The use of PSA as an inte1‘n1t:di'.1te I1‘|Elt‘l\'L’i' at‘ i‘e.~1pti|1.~er: is
`l‘1II'l'l1t’.t“lC‘mt1[)llCt1is3Cl by the fuel
`that PSA levels can he
`ttffected by 1’! vtlricly of I'rtct0I‘:;. Cm‘tiet>+itc1‘citl£< httve the
`capacity to t.'it1.1.‘it'.‘ t'ec|Ltcti0n.~‘. in PSA nnd have been Hhmvii to
`
`[JLJSSCHS palliative and ttntitunior pr0pet'tie.~‘._7 In adtlilioii.
`ttntin11tl1'0{__*en witlulrmvtil cle:trI_v testtlls in PSA t'lCL‘ll]1L"!-'. in
`£tppI‘0.\'lE1ltl[t.:’-l}' 30% of i)atie11ts.“'L’ Thus. some 0!" the PSA
`clntngcs noted in older rep01't5 ill
`the 1ite1':tlLtrc may he
`t1ttt'ibt1tet'l
`tin antizttitliegen \vithdm\va1l or
`the effect 01'
`L'.’t1l‘It3llI'l‘El1[l}" i1Lllttll1l5l't.‘.I'ECl Ct}i‘|lCUSlt‘.‘I‘l)l(l.'-:.
`
`The report that in Harrie preclinical models. some agents.
`result in at PSA decline that is ztttrihutetlttiii1|1ihitioi1oI'PSA
`
`Fmnt the Um'1't~:ri.f_\‘ .-1|-,I"(‘ii.li.')‘i:rriiri, Stir: Fi"tim'r'_w'r.-, C4.‘ Ptil'1l'(“Dttl‘i.§'
`Palmi'H.'m't'mit'tii R¢'_~.'t'.rii‘r'h. Di'I'i'.\'t'mt
`ii!" l-l-'m'iicu’-.l.tiiii.l!t*:'t CH.
`.-'l.rm .=l.-'-
`-I"tU'.
`tl‘N.'
`rt.'J'.r.|'
`..l'mlm.v Hrili:.l'i':i\' Uriit't'r'.rii_t', Btrh'iiiim‘t’. MTJ.
`.S'trhmi.frtJt:" J'm'_r _I'i'. 2rJt)rJ.- m't't*p!t'ul Nrai-'t*:iih.w' if. 3I'lUfJ,
`ti!"
`.-li1l<h'r!.\',\'
`rt';J:'i'm rt*qiw.\'r.\
`in .l':':‘i:' J. Srirrill. MD.
`[i'Htt't'r.\'i.r_\'
`C'r:lfliU"rrfti. Sim .’7r‘tutt'i.n'rt. Ut'_".5T" Citart};rt*.lit'ri.\'r't'¢' ('tmr't'i' Ct‘irit’i'. Milt}
`Di1'i.rm."t~i'n Si.
`.1'm’ Firms‘.
`.'_i'm:
`:|l'ul‘t’t'.llt'J'-.5'{'.'t_ [H E/LINE.‘
`i'imii.l.‘ ,s‘.flltt'i"h:’f9"
`rm-‘nhrititl.HL'.t}'§t’t.r‘H.
`El 3t'}f.li'
`in.‘ .~liHt'ri'.r'tm .\'m'r'.r_‘rt- r:f'C'i’iiit't'tii'
`U7_i'3- .lt‘I'_l5.\’/ll ll/ll UI'l5- l_'?t'l-ll
`
`t'_tli't'{'r.P.l'.‘tg_V.'_
`
`i304
`
`l"tIrther cen-
`2-“.ee1‘eti0n without concurrent cytoloxieity lire;
`fettndcd the ii1terpi'ettiti0n of pt1:atti‘et1tti1ei1t elmnges in PSA
`levels. For cxatntple. the tliscorclnnee l.‘lt3lWCC]"| PSA .\'tl|)|)I'eH—
`sinn ttntl t.l]Iilll.|lTIOi' ttetivity 11:1.-s been reported with sttrztrniii
`in preeliniettl in vitro and in viva I1mtlels.'” However. these
`ltll)()l'fll0t'y 0l1~;e:‘vtt1iei1:<, I'epot'ted in :1 single ptlblieritian. are
`of unkntiwit L'llttltZ£ll sigtiifieztiiec. The rcstills of ‘.1
`large.
`plt'tc‘clJo—cnnti'o|led mttltiecnter I‘nt1drami?.ct'I trial eoi1i]3t1:‘i11g
`sttrntnin plus |1yt'J1'0et1t‘tis0iie tn plztcehu plus l1yt'|me0I'tisnne
`have been recently rep0rtct'l," The ptit'|Jm;e of this 'tII1£Il_\,’!~'lh'
`was to determine whct|1e1' p0!-illI'r?EllIllt?I‘Il
`(lt.‘L'llltL‘.‘-
`in PSA
`were Ll!-§.‘iUCiillt3'tl with eliiiieul
`l‘11I.:‘El!-iLtI'L‘-R oi’
`itnpI'm-ement.
`WCK1026
`Page 1
`Journal at‘ Clinical Oncology, Val 19, Na 5 [March I}, 2001: pp i 304-1 31 l
`
`WCK1026
`Page 1
`
`
`
`PSA DECLINE AS MARKER OF OUTCOME
`
`including survival. time to progression. and duration of pain
`control for patients treated on this trial, the largest prospec-
`tive phase III trial to date of systemic therapy for HRPC.
`
`PATIENTS AND METHODS
`
`EIrgfbir'r'I_v Cr‘irer'io rtrrrl Treorrrrenr Plmr
`
`Eligible patients had hislologically conlirrned adentrcarcirtonta of the
`prostate. with painful bone metastases requiring a stable chronic
`negirnen of opioid analgesics. The details of study design, eligibility
`criteria. and treatment regimen have been reported previously." in
`brief. eligible patients were stratilied by PSA level [5 Hit) ng.’nrL. >-
`l'l}lJ ngi'rnl_.) and presence or absence of soft-tissue metastases and were
`randomized in a double—blind fashion to receive surnmin plus hydro~
`cortisone or placebo plus hydroconisone. Tire therapy received by each
`patient was identified tunblindedi only in the event of progressive
`disease or dose-limiting toxicity {DLT}. Progressing patients found to
`be receiving suramin were withdrawn from the study and observed for
`survival. Patients receiving placebo were eligible to enter cross-over
`and receive open~labe| surarnin on the same 'r'8—rlay regimen. PSA
`levels were measured weekly during treatment.
`their monthly after
`tt'eat1l1crtt ended. Meastrrable lesions were assessed at baseline. at week
`I3, and then every 3 months. Opioid analgesic dosage was corrtintt—
`ottsly adjusted as clinically indicated. Toxicity was graded according to
`the Cancer and Leukemia Group B expanded common toxicityi criteria.
`If grade 3 or 4 toxicity occurred. dosing was interrtrpted until
`the
`toxicity resolved to grade I’. or baseline. Patients who experienced any
`persttttcrrt (3 weeks or more": or recurrent grade 3 or 4 torticity without
`significant antittrmor response were considered to have reached DLT‘.
`and tr'eatment was discontinued permanently.
`
`Resporrse Cr'irer'ict
`
`Pain and opioid analgesic use were the primary indicators of
`response. Each night, patients scored their worst pain over the prior 24
`hours {the daily worst pain} from 0 (rib paint to 10 {pain as bad as you
`can imagine) as part of the Brief Pain inventory. ‘Each night. patients
`also recorded their daily opioid analgesic use. which was subsequcntl y
`converted to morphine equiva|enLs. Pain response was prospectively
`defined before unblinding of the data and was achieved when. for a
`rniriintum of 3 consecutive weeks. pain decreased ?3: three points from
`baseline while opioid analgesic use either decreased or remained stable
`t < |5% increase], opioid analgesic use decreased 2 33% from
`baseline while pain either decrertsed or remained stable ( C I! point
`increase}. or both of these. For patients with baseline pain scores 33: 2
`but
`less than 3. a r'edLIctiot1 to [1 was required to achieve response.
`Similarly. for patients with baseline daily opioid analgesic use a 5 mg
`but less than 15 mg [rnorphine equivalents). a decrease by 5 mg was
`required to achieve pain response.
`Prior reports“ have suggested that posttherapy declines in PSA of
`Z 50% correlate with improved survival. For this reason. patients were
`retrospectively categorized into one of two categories on the basis of
`whether a greater than 5D'y"r- decline from baseline that lasted 2 23 days
`was achieved t for ptttposes of this report.
`termed a PSA declinel.
`Perlirrmairce status was measured weekly with the Validated Revised
`Rand Functional Lirnitations Scale [RRFLSL which captured eaclt
`patier1t's se|i'—assessrnent of everyday activities such as selllcare and
`mobility. ranging from it score of El (least functional inipairlnentl to 40
`{most
`functional
`impairment). as well as by physician—deterrnirted
`Karnofsky Perforntance Status tKPS).
`
`Dr’.ren.re Pr‘rrgr‘e.rsr'nrr
`
`Objective disease progression was defined as an increase in size of
`rricasurable lesions. developrncnl of new osseous lesions. new urinary
`outflow obstruction .-accorrdrny to tumor that required intcrverrlion. new
`malignant pleural effusion. or new spinal cord compression.
`In all
`patients.
`regardless of tlteir pain response status, demonstration of
`objective disease progression was considered evidence of treatment
`liailtrrc and resulted in anblinding of treatment. Pain [subjectivct
`progression was prospectively defined its a
`two-point
`incr'eas'«e
`in
`weekly average of the daily worst pain score or at greater than |5‘}in
`incrsrsc in weekly average of the total daily opioid analgesic intake.
`each with deterioration from lzrascline in RRFLS perforrnance status by
`2 eight points. An accornprrnyirlg decline in RRITLS score was
`mandated to make disease progression defined by pain. opioid require-
`ments. or both more rigorous and therefore more like] y to be clinically
`significant. Treatment assignment was unblinded when either disease
`progression or DLT occurred. and patients found to be on placebo were
`offered open—1rtbel surarnirt. A rninimtrm of 6 weeks on study was
`required before treatment could be unblinded. Unblindiog for disease
`progression before the 6-week nrarlt required study chair approval.
`
`S tarr'str‘c(o' Corr.vr':ler'rrrr'orr.s'
`
`time to pain
`The objective progression-l'ree survival (OPFSl time.
`progression (TTPP), and overall survival time were rneasttred l'r'ont the
`first day of treatnrerrt with the l-'iap|an—Meier' method.” All analyses
`were performed on an irtlent—lo—treat basis. ’I'hus. patients who pro-
`gressed on placebo plus hydrocortisone and who were crossed over to
`sttramin plus hydrncortisone were included in the analysis.
`To evaluate the association between PSA decline and other measures
`of outcome.
`the Kaplan—Meier method was used to compare. overall
`and within each trcatrnent group. the trends for three end points ICJPFS.
`TTPP, and overall survival} for patients with and without a PSA decline
`of 2 5(l%. All analyses were performctl with the Landmurlt Metlrod.”
`‘Three different 1andt't1r1t‘ks were used: week (1. week 9. and week |"_’.
`Patients whose survival was shorter than the landmark point were
`excluded from analysis. This rnellrod has been used as 11 means of
`reducing the inherent bias of ttssessing survival as a
`function of
`response.
`The effect of other variables on the end points was evaluated by
`fitting a Cox model with that variable.” The variables considered were
`two pretreatment stratilication vttt‘iables (PSA level 5 I00 ngJ'n1L r ‘:>
`I00 ngr'mL and presence or absence of nrertstrrable disease} and size of
`treating center. (Study centers were assigned to one of three categories
`on the basis of accrual. with the goal of accotrrrrin g for differing levels
`of farniliarity and expertise with suramin adnrirristrntion that would
`nttltrrully evolve with increasing cnrollntenl. These arbitrary categories
`were [:1] up to eight patients enrolled. [b] 9 to 23 patients enrolled. and
`[c] 24 or more patients enrolled.) In addition. age. race. baseline pain
`nredicttliun requirement. RRFLS performance stattrs. KPS. alkaline
`phosphatase. hemoglobin. and lactate dehydtogcrtase values were
`considered. For each end point. the variable was first entered linearly
`and their twherc applicable). entered as a group of tcrtiles. Models were
`fitted using the variable alone and the variable in the presence of the
`treatment group. Any variable with a P value of less than .lD was
`considered a possible t:on|'oundcr' for the end point. Where the linear
`and the tertile were both significant. the linear fonn in the nrullivariate
`model was used. Once it list of candidate confotrnders was established.
`:1 Cort model was fitted with these I:orrl‘our'rders and PSA r'esponsc."’
`
`WCK1026
`Page 2
`
`WCK1026
`Page 2
`
`
`
`i306
`
`SMALL ET AL
`
`For the ccttltbitted prrprrltitiort tsuramin plus lrydr*oco1'ti.<ntte and placebo
`plus |'ry(lt'ncor‘tisorte]. Inotlels were also titted with and without a term
`for treatment group.
`
`RESULTS
`
`I996. 460 patients
`From February I994 to December
`entered the study: 229 in the sttraniin pltts hydt'ocortisone
`group and 23 I
`in the placebo plus hytlrocortisone group. All
`patients enrolled were included in an intent-to—treat analy-
`sis. with the exception ol' two patients (one on eaclt artrtl
`who were rantlomizetl but were founel to be ineligible zurct
`did not receive treatrrrettt. The two groups of patients were
`balanced with regard to age.
`race. baseline pain score,
`baseline daily opioid artalgcsic requirernertts. RRFLS per-
`formance stattrs. KPS. site ofdisease (bone only 1* bone plus
`soft
`tissue). PSA level, hemoglobin level. anti prior hor-
`monal
`therapy. as |'It'cVi0lltI-'ly described. D1’ 230 patients
`rartdontized to placebo plus ltydroco1'tisone.
`l64 (71.3%)
`crossed over to receive strramitt al"t'er progressing.
`Of 369 sttramiu patients. the pcrccrttage of patierits with
`a 2 5t)% decrease in PSA lasting at least 28 days. was 23%.
`39%. and 45% with a 6~week. 9-week. and 13-week
`
`|antlniar'k, respective-ly. The respective percerttttge ol‘ 331
`patients receiving placebo with a greater than 509% PSA
`decline was 5%, 21%, and 30%. For all 460 patients
`enrolled,
`the percentage with a PSA decline at
`the three
`l:Itt"rLlt‘I1c'll'lx'1i was
`If-l%. 31%, and 38%. Duration of pain
`t'esponse,
`time to pt'og1'essiott, and overall survival in the
`two treatmettt groups have been reported elsewltere.”
`Disease progression oectrrred in 376 (82%) of the 458
`
`patients and was doctmtertletl on the basis of objective
`evidence in approximately 35% of disease pt'ogt'essors.
`Subjective evidence of disease progression (an ittcrease in
`pain. analgesic use. 01' both. combined with t'letet'iuratiort in
`RRFLS pcrtbrnrtmce status} accounted For [299 and l7°7r- of
`
`patients with pt'ogt'essive disease in the suramin and placebo
`grottps.
`respectively. The details of patterns ol’ disease
`pr'ogression have been reported elsewhere."
`
`/ls.\‘r.Ic."mr'r;rr QIPSA Declitte l"l/ill: Ot'et'rrll .S'ttt‘t'r'r'ttl
`
`When the entire cohort (sttr'amin and placebo patiertts) is
`considered. a durable decline in PSA was found to be
`
`the 6~week
`associated with improved median stlrvival at
`|andmar|~; (563 1' 325 days. P = .00l9l. 9-week lill‘lLll1tllt‘l-t
`(596 1' 344 days. P = .0009}. and I2-week landmark (597 t-'
`374 days. P = .003}. A decline in PSA in patients
`rantlortiized to receive sttrttmirt was also associatted with an
`increased median survival at the 6-week li1l]Clll‘J£1t'l«'. (532 I-'
`
`3 I2 days, P I .0033). 9-week larttlntat'k [532 r 333 (lays. P
`= .0028). and I2-week laridrmtrk {S96 1-’ 392 days. P =
`
`Table l. Association Between PSA Decline [2 50% decrease in P5-A
`lusting at least 28 days] at Landmarks and Overall Survival Showing the
`P Value From Log-Rani: Test lunivctrinte analysis]
`Median
`[dcrysl
`
`9i":
`
`P {log-rctnltl
`
`Population
`landmark
`
`PEA Decline
`
`N
`
`DUI 9
`
`Combined
`6 weeks
`
`l2 weeks
`
`St.-rornin
`6 weeks
`
`9 weeks
`
`12 weeks
`
`Placebo
`6 weal-is
`
`9 weeks
`
`l 2 weeks
`\
`
`All
`Decline
`No decline
`All
`Decline
`No decline
`All
`Decline
`No decline
`
`All
`Decline
`No decline
`All
`Decline
`Ne decline
`All
`Decline
`No decline
`
`All
`Decline
`
`No decline _”
`All
`Decline
`No decline
`All
`Decline
`No decline
`
`346
`50
`296
`278
`35
`I93
`209
`80
`129
`
`l75
`Ill
`i311
`l53
`59
`911
`I l3
`53
`:55
`
`l7l
`9
`
`let?
`l 25
`26
`9?
`91
`'2?
`64
`
`.003). In the placebo plus l1ytl1‘ocot'tisone—tr‘eatetl patients.
`there was a trend toward increased survival
`in patients
`exhibiting a, decline in PSA ["600 t-’ 337 days at the 6-week
`landmark. 612 1' 355 days at the 9-week l}1t‘1(ltTI£1I'l{. and 601
`r 368 clays at the I2-week landmarlct. However, these trends
`
`did not reach statistical signiticurtce, perhaps because of the
`small ntrmber ol patients with a PSA decline in the
`p|rtccbo—treated patients {Table l l.
`Table 2 shows the P values used to screen the variables
`
`(potential conl’our1t]crst for use in the multivariate analysis.
`Only two variables were excluded from the ntullivariatc
`analysis on the basis ol‘ their performance in this screen:
`baseline pain mcdicatiort and race.
`A mtr|tivariaI.e fit of the eFl’ect of PSA decline on overall
`survival. adjtrsted for three strtttificatiort variables (center
`size, baseline PSA. and measurable disease) and the
`
`screened cot1t’otrttt'|e1‘s. did not strbstantially alter the obset'—
`vation that a PSA resportse at any of the lantlniarks resulted
`WCK1026
`Page 3
`
`WCK1026
`Page 3
`
`
`
`PSA DECLINE AS MARKER OF OUTCOME
`
`Table 2.
`
`P Values of Possible Canfaunders and Their Effect on Overall
`Survival From Day 0
`
`No response
`—‘—' Response
`
`Predictor
`
`Measurable disease
`Baseline PE-A
`Size oi center‘
`
`Agel
`Alkaline phosphatase
`Baseline KP5
`
`Hemoglobin
`Lactate clehydrogenase
`RRFI. total score
`Baseline pain medicaliani
`Race‘l’
`
`Linear
`
`«z. _0001
`.0181
`.0676
`
`-
`
`.1752
`' .0001
`. .0001
`
`'. .0001
`.0001
`.0001
`.6196
`.9235
`
`"Defined as small {accrual oi one to eight patlenlsl, medium [accrual ai nine
`to 23 patients}, or large [accrual oi 241 or more patients].
`i
`‘llerlile larrn used in multivariate analysis.
`¥Nol included as variables in multivariate analysis.
`
`in an improvement in median overall survival. both in the
`entire group as well as the rsuramin group (Table 3}. For
`i|1ttsLt'ative purposes. overall suwival of the entire cohort 01‘
`patients as at
`function of PSA decline at
`the 9-week
`luntlmark is shown in Fig I.
`
`A.\'.mt'r'rm'on of PSA Dt=fr_'H.'tt'.‘ Wr'.'l'i {_3PF.S'
`_
`_
`_
`,
`_
`_
`A ClCCltl‘IE l1‘t PSA was 11150 l0lIItL1 it] be [lSSDCl'dle(.l Wlllt
`improved median OPFS in the entire group at all
`three
`lttitdtttttdut (I70 1’ 88 days at
`the 6-week landmark. P =
`.0027: 183 1' 96 days at the 9-week landmark. P < .0000];
`and 193 1' 122 dztys at the 12-vtlieelt
`|ztnt1|na|'k, P < .0001}.
`In the 1-‘uramin plus hydroeortisone group-.:1nd placebo plus
`hydrocortisone group, adecline in PSA w't‘t.‘i‘i't!:‘-I-}0t’.‘ltllE(_l with
`the OPFS when the 9- and 12-week Iattdiharks were used
`but did not reach statistical sigitilieattce wltep the 6-week
`l£tI1(lt'l‘l:1l'1{ was used (P = .0662 for
`the suramin plus
`
`.
`.
`600
`400
`Overall survival {Days}
`
`F191. Overall survival [days] by PSA decline at the 9-week landmark [all
`Pufie,-.g5}_ Median [clays]: response = 596 days; no response = 344 days
`lP=.E|OD91»
`
`hytlrocortisnne group: P = .0582 for the placebo plus.
`l1y(lt‘DC{}I'llSD1]t‘) group) (Table 4).
`As with the analysis at‘ vtlriables‘ (possible CUI1f0Llt1(lEl‘Sl
`of the 1‘e1atian.~;|1ip between PSA decline and overall
`t-nir-
`viva]. only two vmiables tbaselitte pain Inedietttiott and
`race) were excluded from the multivariate analysis of the
`
`Population
`Combined
`
`Table 3. Multivariate Analysis at the Association Eetween FSA Decline at Landmarks and Overall Survival, OPF5. and ITPP With the Cox Model
`Chrarull Survival
`OFFS
`TTPP
`Landmark
`P
`HR
`P [Cox]
`HR
`F lCax]
`HR
`lweeltsl
`95% Cl
`95% C!
`95% CI
`.0037
`2.41
`.0045
`1.20
`.0025
`1.63
`6
`1.14-2.311
`1.18-2.46
`13341.35
`.0075
`2.00
`-1 .0001
`1.96
`-1 .0024
`1.61
`9
`1.18-2.18
`1.42-2.72
`1.20-3.33
`.0515
`1.72
`-:
`.0001
`2.19
`-1. .0068
`1.59
`12
`1.14-2.23
`1.52-3.17
`1.00-2.97
`.019?
`2.46
`.1088
`1.115
`.0031
`1.90
`6
`1.24-2.91
`0.92-2.27
`1 .15-5.26
`.0135
`2.58
`.0309
`1.62
`.0009
`1.99
`9
`1.33-2.99
`1.05-2.52
`1.22-5.48
`.0350
`2.219
`.0010
`2.36
`.0002
`2.37’
`12
`1.50-3.75
`1 .41—3.93
`1.03’-5.80
`.4162‘?
`1.50
`.1065
`1.9?
`.2883
`1.57
`6
`0.68-3.64
`0.8?’-4.47
`0.50-4.46
`.2867
`1.49
`.0016
`2.41
`.0541
`1.69
`9
`0.99-2.82‘
`1.40-4.16
`0.71-3.13
`.9225
`1.04
`.01-42
`2.17
`.2850
`1.36
`12
`037-2111
`1.1?-41.03
`0.47-2.30
`NOTE. Table shows the hazards ratio with 95% conlidence interval and P value {adjusted lor potential canlounclers, including the three stratilicatiart variablesl.
`Abbreviations: HR, hazards ratio; CI, confidence interval.
`
`Suramin
`
`Placebo
`
`WCK1026
`Page 4
`
`WCK1026
`Page 4
`
`
`
`I308
`
`PSA Decline
`
`N
`
`Table 4. Association fietvveen PSA Decline oi Landmarks and Time to
`Obieciive Progression Showing the P Value From Log-llonk Test
`.l\I‘I.eclicIr1
`Popululion
`Loridmorlt
`[daysl
`
`'36
`
`P [log-rnnltl
`
`.0027
`
`Combined
`
`6 weeks
`
`9 weeks
`
`I2 weeks
`
`Placebo
`6 weeks
`
`9 weeks
`
`12 weeks
`
`All
`Decline
`No decline
`All
`Decline
`No decline
`All
`Decline
`No decline
`
`All
`Decline
`No decline
`All
`Decline
`No decline
`All
`Decline
`No decline
`
`All
`Decline
`No decline
`All
`Decline
`No decline
`All
`Decline
`No decline
`
`relationship bctwccn PSA decline and OPFS (data not
`shown). In a multivariate analysis, PSA dcclinc remained
`predictive of median OPFS at all
`three landmarks for the
`combined population. and at the 9- and 12-wcck landmarks
`
`for thc suramin and placebo groups (Table 3). A PSA
`decline in the stiramin-treated group was associated with an
`improvement
`in median OPFS at
`the 9-week landmark
`("hazards ratio, L62; P = .03) and at the 12-week landmark
`
`(hazards ratio, 2.36; P = .001). As an cxamplc,prog1'cssion—
`free survival of all patients as a function of PSA decline
`with a 9-week landmark is shown in Fig 2.
`
`Assricr'rm'0ii of PSA Decliiie Writ’: TTPP
`
`A decline in PSA was associated with improved median
`TTPP in the critirc cohort at all three landmarks (358 days
`I’ 184 days at the 6-week landmark. P = 00269; 428 1-‘ [89
`days at the 9-week landmark, P = .0009; and 428 1’ I85
`days at the 12-week landmark, P = .0131). In the surarnin
`group. a decline in PSA was also associated with an
`
`SMALL ET AL
`
`‘I
`
`No response
`1' Response
`
`100
`
`200
`
`300
`
`400
`
`500
`
`Time to Objective Progression (days)
`
`Fig 2. Time to obieclive progression by PSA decline at 9-week landmark
`{all patients]. Median {days}: response = I83 days; no response = 96 days
`IF < .0001.
`\
`
`improved median TTPP at all three landmarks {"358 days v
`269 days at 6 weeks. P = .0208: 392 days v 184 days at 9
`weeks. P = .0038; and 392 days v 184 days at 12 weeks, P
`= .0208). No association between PSA decline and median
`
`TTPP was pbservcd in the placebo group (Table 5}. All
`potential confoundcrs were included in the mLtltiva1'iatc
`
`analysis. In a multivariate analysis, a PSA decline rcmaincd
`predictive of median TTPP at all thrcc landmarks for the
`
`entire group and the stiramiii-ticated group, with hazards
`ratios between l.?2 and 2.4]
`for die entire group and a
`hazards ratio of approximately 2.5 for the suramin group
`(Table 3). For illustrative purposes, TTPP of all patients as
`a function of PSA decline using a 9-week landmark is
`shown in Fig 3.
`
`DISCUSSION
`
`The majority of HRPC patients lack measurable disease,
`and responses in osscous disease are difficull to quantify.
`Consequently, clinically relevant and points such as pain
`control and analgesic use have become more widely uscd.'5""
`WCK1026
`Page 5
`
`WCK1026
`Page 5
`
`
`
`PSA DECLINE AS MARKER OF OUTCOME
`
`Table 5. Association Between PSA Decline at Landmarks and Time to
`Pain Progression Showing the P Value From Log-Rank Test
`Median
`Population
`l.CIF1Clt'nC:rl<
`inlays]
`
`P llog-ranlxl
`
`FSA Decline
`
`N
`
`D0269
`
`Combined
`6 weeks
`
`9 weeits
`
`12 weeks
`
`Suramin
`6 weelcs
`
`9 weeks
`
`l 2 weelzs
`
`Placebo
`6 weeks
`
`9 weeks
`
`12 weelts
`
`All
`Decline
`No decline
`All
`Decline
`Na decline
`All
`Decline
`No clecline
`
`All
`Decline
`Na decline
`All
`Decline
`No decline
`All
`Decline
`No decline
`
`All
`Decline
`Na decline
`All
`Decline
`No clecline
`All
`Decline
`No decline
`
`G5
`95
`
`20
`"B0
`
`30
`70
`
`2.5
`9?
`89
`2?
`I52
`
`l7l
`lo‘?
`
`395'
`l34
`
`399
`135
`
`NOTE. Six patients did not have Fallow-lup pain data available and Iherelore
`cannot be included in this analysis.
`ll,
`
`the asscssi1tcnt' ot’ syniptomatic "end points as a
`However.
`incasure of tlicrapctitic benefit can be associated with substan-
`tial ntcthodologic difficultics. PSA declines liavc been rc-
`poitctl. often in lieu at" other more cstablislted markers of
`1‘ESpt)l1Ht:. The correlation of PSA changes with clinically
`signilicant end points continues to be dcbtitcd.” ln I.his slutty.
`a roughly two-l"old iiicicasc in the PSA response proportion in
`the sttrztmin arm L‘ompaI‘c(l with placebo mit'rot'c(l other mark-
`ers of response that l"avorcd suramin. including pain response.
`risk of progression. antl durability of pain responses." As a
`consequence, a inorc cm-el"ul
`atta|_v.~;is of the rclationsltip
`between PSA decline and several clinical outcomes was
`
`l|1'l(lEI'l.il.l\i.E|'l. This exploratory analysis has shown that in this
`group of patients. a postthcrttpy decline in PSA of 2 50%
`lasting 2: 28 days is associated not only with prolonged median
`overall survival, bill also with prolonged median OPFS and
`proloitgcd median TFPP in men with symptomatic HRPC.
`"this association held in the entiri: study colioit ovcrtlll. incle-
`
`No response
`Response
`
`"‘i—
`0
`
`200
`
`300
`
`400
`
`Time to pain progression {Days}
`
`Fig 3. Time to pain progression by PSA decline at 9-week landmark {all
`patients]. Median ldoysl: response = 428; no response = 134 IP = .0009].
`
`pendent of ttcatmcnt. as well as among those patici1tsrct:civing
`suramin trcatrnent. and with tltrec dit’l'ercnt lEt11(lIl1‘:1l‘liS. F1.irtlict'—
`
`mans, this positive association was apparent both on univariale
`and inultivariate analysis.
`In assessing the predictive utility of posttrcatmcnt PSA
`changes. previous rcpons have appropriately l’ocusct:l on the
`|1ardcs1' end point: stn'viva|.3'5 However. ‘correlation of PSA
`declines with other markers of clinical benefit might
`
`strcngtltcti the argument that it PSA dcclinc can serve as an
`iittcmnediatc end point
`in clinical
`trials involving HRPC
`patients.
`in addition to anal_vzing thc rclatiorisliip bctwccri
`PSA decline and survival, this data set‘ permitted a correla-
`
`including
`tion with other clinically important end points.
`median TTPP and OPFS. Criteria for both pain response and
`pain progressioii were prospectively defined before unblind—
`ing and well bclorc any correlation with PSA data was
`undertaken. A significant associatioit between PSA decline
`and both OPFS and TTPP was observed. This association-
`between PSA decline and clinically significant subjcctivc
`
`end points such as pain control suggests that PSA changes
`may have greater i'amifications than generally appreciated.
`WCK1026
`Page 6
`
`WCK1026
`Page 6
`
`
`
`l3l0
`
`SMALL ET AL
`
`To our knowledge, this study represents the largest prospec-
`tive series reported correlating PSA changes with patient-
`derived end points such as pain control. Nevertheless.
`because the overall results of this study did not denrorrstrate
`:1 survival €1dV"i'l1'tl;lgt3 of otre treatment amt over the other‘.
`the conditions l'or surrogacy (of PSA vis—a—vis survival)
`were not tnetz"
`
`Several groups have previously reported on the predictive
`valtre of posttreatment decline in PSA in phase II trials for
`the treatment of HRPC.” Sttch analyses derived from
`uncontrolled experience should be viewed as hypothesis
`generating and require further
`testing in prospectively
`designed randomized trials.
`Investigators from Memorial
`Sloan-Kettering“ and from the University of Michigan]
`have shown that a durable greater than 50% decline in PSA
`
`at an 8~vveek IEi.l1CII1‘l:1I'I\‘ was an independent predictor of
`survival in HRPC patients. Investigators at the University of
`Maryland conducted a similar analysis on data collected on
`I03 patients enrolled onto two phase I and II studies ot‘
`strramin. Although a relationship between PSA declines at 4
`weeks and survival was seen.
`these investigators were
`unable to identify a specific drreshold of PSA decline
`consistently associated with a survival advantage.‘ This
`study may have failed to show such an association because
`of smaller sample size. In addition,
`this series included a
`
`phase I and ll study. which included diflercrrt dosing
`regimens that could conceivably have had diffenent thera-
`peutic effects.
`
`Laboratory studies have strggested that suramin may
`affect PSA levels without necessarily resulting in cell death.
`Prostate cancer cell litres grown in xcnograft models have
`been reported to have reduced expression of PSA mRNA
`withottt associated cell deatlt when treated in vitro with
`suramin.'° This observation has been the basis of recom-
`
`mendations urging caution in the use of PSA as a predictor
`of response in surarrrin-treated patierrts.“ However.
`this
`data set demonstrates a significant relatioasltip between
`PSA decline and clinically relevant outcomes tover'all
`survival.
`time to objective progression, and TTPP} in
`suramin-treated patients. The. results of this analysis suggest
`that preclinical assays designed to evaluate the effects of
`drugs on PSA secretion may not necessarily be clinically
`significant and require proper validation before they can be
`used to interpret results from clinical trials.
`
`In summary. this database. derived from the treatment of
`symptomatic HRPC‘ patients with either suranrirr plus hy-
`drocor1.isorre or placebo plus hydrocortisone on a prospec-
`tive, placel:to~conlrollcd. double-blirrd trial. supports the
`observation that a durable' decline in PSA is associated
`with overall survival. Furthermore. a durable PSA de-
`
`cline. was also associated with other markers of response,
`including TTPP and OPFS. This analysis indicates that
`previously reported data on the in vitro effects of suramin
`on PSA secretion are likely to have limited clinical
`nelevancc and raises a larger qtrestion as to the utility of
`such preclinical assays.
`L
`
`REFERENCES
`
`trials in relapsed
`I. Sclier I-ll. Mazttrndar M. Kelly WK: Clinical
`prostate cancer: Defining the target. J Natl Cancer Inst SB:lfi23-1634.
`I996
`
`1. Kelly WK. Scher HI. Mazunrdar M: Prostate—specilic antigen as
`a rrreastrre of disease outcome in metastatic lrormone—refrt1ctory pros-
`tate cancer. J Clin Oncel
`ll:6(l’r'-fil5. 1993
`
`El. Smith DC. Dttnn RL. Sttawdermatr MS. et at Change in serum
`prostate—spccil'i(: antigen as a marker of response to cytotoxic therapy
`for ltormone-refractory prostate cancer. J Clin Oncol
`lfr:|335—|S43.
`I998
`
`ill! Post—tlIerapy serum
`4. Scher Hl. Kelly WK. Zlrang ZF. et
`pr'ostate—specific antigen level and survival in patients with :1nr|rogen-
`independent prostate cancer. .1 Natl Cancer Inst 9l:24s'l-~25|, I999
`5. Sridlrara R. Eisenberger M. Sinibaltli V. et al: Evaluation ol’
`prostate specific arttigen as a surrogtrtc marker for response ol'l1ormonc
`refractory prostate cancer to suramin therapy. J Clin Oncol
`t3:2944—
`2953. 1995
`
`(1. Dawson NA: Apples and oranges: Building a consensus for
`standtwdized eligibility criteria and end points in prostate cancer
`clinical trials. J Clin Oncol
`|fi:fi39FI-3-‘I05. I993
`?. Tannock I. Crtrspodarowicz M, Mcaltin W. et al: Treatment of
`rrtctastatic prostate cancer with low—dose prcdrtisone: Evaluation of
`pain and quality of life as pragmatic indices of response. J Clin Oncol
`7:590-597. I989
`
`8. Scher HI. Kelly WK: Flutarnitle vvitlrdrtlwtrl syndrome: Its impact
`on clinical trials itr |1orrnone—refractory prostate cancer. J Clirr Onc