V O L U M E 2 2 䡠 N U M B E R 6 䡠 M A R C H 1 5 2 0 0 4
`
`JOURNAL OF CLINICAL ONCOLOGY
`
`O R I G I N A L R E P O R T
`
`Antiandrogen Withdrawal Alone or in Combination
`With Ketoconazole in Androgen-Independent Prostate
`Cancer Patients: A Phase III Trial (CALGB 9583)
`Eric J. Small, Susan Halabi, Nancy A. Dawson, Walter M. Stadler, Brian I. Rini, Joel Picus,
`Preston Gable, Frank M. Torti, Ellen Kaplan, and Nicholas J. Vogelzang
`
`A
`
`B
`
`S
`
`T
`
`R
`
`A
`
`C
`
`T
`
`Purpose
`Antiandrogen withdrawal (AAWD) results in a prostate-specific antigen (PSA) response (decline in PSA
`level of ⱖ 50%) in 15% to 30% of androgen-independent prostate cancer (AiPCa) patients. Thereafter,
`adrenal androgen ablation with agents such as ketoconazole (K) is commonly utilized. The therapeutic
`effect of AAWD alone was compared with simultaneous AAWD and K therapy.
`
`Patients and Methods
`AiPCa patients were randomized to undergo AAWD alone (n ⫽ 132), or together with K (400 mg orally
`[po] tid) and hydrocortisone (30 mg po each morning, 10 mg po each evening; n ⫽ 128). Patients who
`developed progressive disease after AAWD alone were eligible for deferred treatment with K.
`
`Results
`Eleven percent of patients undergoing AAWD alone had a PSA response, compared to 27% of patients
`who underwent AAWD and simultaneous K (P ⫽ .0002). Objective responses were observed in 2% of
`patients treated with AAWD alone compared to 20% in patients treated with AAWD/K (P ⫽ .02). There
`was no difference in survival. PSA and objective responses were observed in 32% and 7%, respectively,
`of patients receiving deferred K, and were more common in patients with prior AAWD response.
`Treatment with K was well tolerated, and resulted in a decline in adrenal androgen levels, which rose at
`the time of disease progression.
`
`Conclusion
`K has modest activity in AiPCa patients, while AAWD alone has minimal activity. Adrenal androgen levels
`fall with treatment with K and then climb at the time of progression, suggesting that progressive disease
`while on K may be due to tachyphylaxis to the adrenolytic properties of K.
`
`J Clin Oncol 22:1025-1033.
`
`INTRODUCTION
`
`Prostate cancer is the most common cancer in
`men and will account for more than 30,000
`deaths in 2003 [1]. The vast majority of deaths
`are due to the development of metastatic dis-
`ease unresponsive to androgen deprivation.
`Androgen deprivation, with gonadal andro-
`gen suppression with or without antiandro-
`gen, has been the standard of care in pa-
`tients with metastatic disease since the
`1940s [2] and is being increasingly utilized
`in patients with less advanced disease [3].
`The benefits of adding an antiandrogen
`such as flutamide or bicalutamide to go-
`
`nadal androgen suppression at the time of
`initiating androgen deprivation (early use)
`are modest at best [4]. The late addition of
`an antiandrogen after initial failure of an-
`drogen deprivation (late use) seems to result
`in prostate-specific antigen (PSA) declines,
`and in some cases, objective responses [5].
`Thus, at one point or another, most patients
`with advanced prostate cancer will be
`treated with an antiandrogen. Despite an-
`drogen deprivation, including the use of an
`antiandrogen, most patients will experience
`disease progression. For patients with pro-
`gressive disease, despite androgen depriva-
`tion, withdrawal of antiandrogen has been
`
`From the University of California San
`Francisco, San Francisco, and US Naval
`Medical Center, University of California
`San Diego, San Diego, CA; Cancer and
`Leukemia Group B Statistical Center,
`Duke University Medical Center,
`Durham, and Wake Forest Comprehen-
`sive Cancer Center, Winston-Salem,
`NC; Greenebaum Cancer Center,
`University of Maryland, Baltimore, MD;
`University of Chicago Medical Center,
`Chicago, IL; and Washington Univer-
`sity Barnard Cancer Center, St
`Louis, MO.
`
`Submitted June 10, 2003; accepted
`January 6, 2004.
`
`Supported by grants CA60138 (E.J.S. and
`B.I.R.), CA47577 (S.H. and E.K.),
`CA31983 (N.A.D.), CA41287 (W.M.S. and
`N.J.V.), CA77440 (J.P.), CA11789 (P.G.),
`and CA03927 (F.M.T.). The research for
`Cancer and Leukemia Group B Trial 9583
`was supported, in part, by grants from
`the National Cancer Institute (CA31946)
`to the Cancer and Leukemia Group B
`(Richard L. Schilsky, MD, Chairman). The
`research was also supported by Janssen
`Pharmaceutica Products, LP.
`
`Presented in part at the American Soci-
`ety of Clinical Oncology annual meeting,
`San Francisco, CA, May 12-15, 2001.
`
`The contents of this manuscript are
`solely the responsibility of the authors
`and do not necessarily represent
`the official views of the National
`Cancer Institute.
`
`Authors’ disclosures of potential con-
`flicts of interest are found at the end of
`this article.
`
`Address reprint requests to Eric J.
`Small, MD, UCSF Comprehensive Can-
`cer Center, University of California
`San Francisco, 1600 Divisadero St,
`Room A-718, San Francisco, CA 94115;
`e-mail: smalle@medicine.ucsf.edu.
`
`0732-183X/04/2206-1025/$20.00
`
`DOI: 10.1200/JCO.2004.06.037
`
`Downloaded from jco.ascopubs.org on July 20, 2016. For personal use only. No other uses without permission.
`Copyright © 2004 American Society of Clinical Oncology.
`
`1025
`
`
`
`
`JANSSEN EXHIBIT 2063
`Mylan v. Janssen IPR2016-01332
`
`

`

`Small et al
`
`reported to result in a decline in PSA level of ⱖ 50% in 15%
`to 30% of patients [6-8]. The mechanism of this phenome-
`non has been attributed to alterations in the androgen sig-
`naling cascade, including mutations in the androgen recep-
`tor (AR), resulting in the antiandrogen behaving as an
`activator, not inhibitor, of the AR [9].
`The duration of decline in PSA observed with antian-
`drogen withdrawal (AAWD) is brief, with a median dura-
`tion of 3.5 to 5.0 months [6-8], after which further therapy
`is generally required. The mechanism by which prostate
`cancer patients develop disease progression after AAWD is
`not understood, but it has been postulated that persistence
`of a clone of cells with partial or full sensitivity to testoster-
`one might be provided a growth advantage by androgen
`produced by the adrenal glands. If this were the case, it could be
`anticipated that adrenal androgen suppression would demon-
`strate some anticancer activity in this setting. An early report
`suggested that the addition of aminoglutethimide (an adrenal
`steroid synthesis inhibitor) at the time of AAWD increased the
`percentage of patients with a decline in PSA over that which
`would be expected with antiandrogen alone [10].
`Ketoconazole is an azole antifungal agent which exerts
`its clinical effect through the inhibition of cytochrome P450
`14a-demethylase, a catalyst of the conversion of lanosterol
`to cholesterol. Ketoconazole has been in clinical use as an
`antifungal agent for more than 20 years. In its initial use as
`an antifungal agent, it was discovered that a proportion of
`men who used ketoconazole developed painful gynecomas-
`tia, which was later found to be due to the suppression of
`testicular and adrenal androgen production, and it was
`postulated that this adverse effect could be useful for pros-
`tate cancer therapy. Several trials have evaluated the use of
`ketoconazole in patients with androgen-independent pros-
`tate cancer (AiPCa), though most predated the use of PSA
`or an understanding of the AAWD syndrome [11-14]. Most
`recently, the use of ketoconazole after AAWD was reported
`in a trial of 48 patients, 30 (63%) of whom demonstrated a
`ⱖ 50% decline in PSA response. The decline of ⱖ 50% in
`PSA was comparable in patients who had initially re-
`sponded to AAWD and in those who had not initially re-
`sponded to AAWD at 65% v 40%, respectively. Toxicity was
`largely mild in these patients, consisting of grade 1 or 2
`nausea, fatigue, edema, hepatoxicity, and rash [15].
`It was hypothesized that the simultaneous addition of
`ketoconazole to AAWD would have additive anticancer
`activity, simultaneously targeting the stimulating effects of
`antiandrogen and adrenal androgens. Hence, the principal
`goals of this study were to prospectively evaluate and com-
`pare the effect of AAWD alone on PSA levels, compared
`with simultaneous AAWD and ketoconazole therapy. In
`addition, the effect of ketoconazole used in patients whose
`cancer had progressed after AAWD was also evaluated.
`
`PATIENTS AND METHODS
`
`Patients
`Eligible patients had histologically confirmed adeno-
`carcinoma of the prostate with progressive metastatic dis-
`ease, as defined below, despite anorchid testosterone levels
`(⬍ 50 ng/mL). Androgen deprivation therapy was required
`to include at least 4 weeks of ongoing therapy with an
`antiandrogen (flutamide, bicalutamide, or nilutamide).
`Ongoing gonadal androgen ablation with a luteinizing
`hormone-releasing hormone analog or orchiectomy was
`required. For patients with measurable disease, progression
`was defined as a greater than 25% increase in the sum of the
`products of the perpendicular diameters of all measurable
`lesions. For patients with “bone only” disease, a PSA greater
`than 5 ng/mL, which had risen from baseline on at least two
`successive occasions at least 4 weeks apart was required.
`Patients were required to have had metastatic disease dem-
`onstrated on imaging at some point during their history,
`but were not required to have demonstrated metastases on
`imaging at the time of enrollment. Patients were excluded if
`they had received prior chemotherapy, immunotherapy,
`experimental therapy, or prior treatment with ketocon-
`azole, aminoglutethimide, or corticosteroids if they had a
`Cancer and Leukemia Group B (CALGB) performance
`status of more than 2, total bilirubin level greater than 1.5⫻
`the upper levels of normal (ULN), or serum glutamic-
`oxaloacetic transaminase level greater than 3⫻ ULN. Be-
`cause of potential interactions with ketoconazole, no ongo-
`ing or concurrent use of
`terfenadine, astemizole, or
`cisapride was allowed. All participants signed an institu-
`tional review board–approved, protocol-specific, informed
`consent form in accordance with federal and institutional
`guidelines.
`
`Treatment
`After registration, patients were randomized to AAWD
`alone or AAWD ⫹ ketoconazole by the CALGB Statistical
`Center. This study was neither blinded nor placebo con-
`trolled. A conventional stratified random blocks design was
`used [16]. That is, within each stratum, patients were as-
`signed to the available two treatments in blocks of treatment
`assignments so that an equal number of patients was as-
`signed to each of the two treatment arms within each block.
`Randomization was stratified by four features that could
`potentially affect the likelihood of response, both to AAWD
`and/or ketoconazole therapy: (1) prior therapy with flut-
`amide, bicalutamide, or nilutamide; (2) continuous or in-
`termittent treatment; (3) initial or delayed antiandrogen
`therapy; and (4) imaging studies positive or negative for
`metastatic disease.
`Patient registration, subsequent randomization and
`data collection were managed by the CALGB Statistical
`Center. Furthermore, data quality was ensured by careful
`
`1026
`
`JOURNAL OF CLINICAL ONCOLOGY
`
`Downloaded from jco.ascopubs.org on July 20, 2016. For personal use only. No other uses without permission.
`Copyright © 2004 American Society of Clinical Oncology. All rights reserved.
`
`

`

`AAWD and Ketoconazole for Prostate Cancer
`
`review of all data at the CALGB Statistical Center and by the
`study chairperson. Patients enrolled on the AAWD ⫹ keto-
`conazole arm received ketoconazole 400 mg tid po plus
`hydrocortisone 40 mg/d po (30 mg each morning and 10
`mg every night) continuously until disease progression or
`unacceptable toxicity, as described below. Patients random-
`ized to the AAWD-alone arm were required to cross over to
`treatment with ketoconazole on disease progression (see
`Response and Progression Criteria).
`Eligible patients were evaluated with a medical history
`and physical examination at study entry and monthly there-
`after. In addition to a complete medical history and physical
`examination at each visit, patients were evaluated for ad-
`verse events. No formal quality-of-life or pain assessment
`was undertaken. CBC, PSA, total bilirubin, alkaline phos-
`phatase, asparate transaminase, creatinine, glucose, and lac-
`tate dehydrogenase (LDH) were checked at baseline and
`then monthly. An endocrine panel including androstendi-
`one, dehydroepiandrosterone sulfate (DHEAS), dehydro-
`epiandrosterone (DHEA), and testosterone was obtained at
`baseline, at 1 month after starting therapy, at 3 months after
`starting therapy, and at the time of disease progression.
`Blood samples were obtained between 8:00 AM and 10:00 AM.
`Plasma was isolated, frozen, and shipped for analysis at a
`central laboratory to determine androstendione, DHEAS,
`DHEA, and testosterone levels. A bone scan and computed
`tomography scan of the abdomen and pelvis were obtained
`at baseline. If imaging studies were positive at baseline, they
`were repeated every 3 months.
`Replacement doses of hydrocortisone were continued
`as long as the patient was receiving ketoconazole. When the
`patient was no longer receiving ketoconazole, hydrocorti-
`sone was tapered by 5 mg every 3 days until completely
`discontinued. Antacids, H-2 blockers, and proton pump
`inhibitors were avoided but not explicitly prohibited. This
`study did not mandate that ketoconazole be taken on an
`empty stomach or with acidifying procedures. At each visit,
`toxicity was graded according to the National Cancer Insti-
`tute common toxicity criteria (CTC, version 2.0) and re-
`corded. In the event of grade 3 or higher hepatotoxicity or
`symptomatic peptic ulcer or gastritis, patients were re-
`moved from protocol treatment. Antiemetics other than
`corticosteroids were permitted. If grade 2 or 3 nausea per-
`sisted despite these measures, the patient was removed from
`therapy. Patients developing other grade 3 or higher toxicity
`had treatment held until toxicity resolved to grade 1 or
`better. Any patient developing grade 4 toxicity or grade 3
`toxicity persisting for longer than 4 weeks, except as out-
`lined previously, was removed from protocol treatment.
`
`Response and Progression Criteria
`This study was launched before publication of the Re-
`sponse Evaluation Criteria in Solid Tumors criteria or the
`PSA Consensus Criteria [17], so that a composite end point
`
`combining PSA changes with imaging changes was utilized.
`Therefore, for patients with measurable disease, a partial
`response was defined as a ⱖ 50% decrease in the sum of the
`products of the perpendicular diameters of all measurable
`lesions, together with a decline in PSA of ⱖ 75%, measured
`at least twice at least 2 weeks apart, whereas a complete
`response was complete resolution of all visible disease, and
`normalization of PSA on at least two occasions at least 2
`weeks apart. A complete response in patients with bone-
`only disease was defined as complete normalization of bone
`scan, again with a normalization of PSA, while a partial
`response in bone-only patients requires a greater than 75%
`decline in PSA with no new lesions on bone scan. Addition-
`ally, all patients had monitoring of PSA levels, and the
`percentage of patients with a decline in PSA of ⱖ 50%,
`documented on at least two successive occasions, at least 4
`weeks apart, was calculated per the PSA Consensus Criteria
`[17]. An intent-to-treat analysis was utilized. Patients who
`received less than 2 months of therapy for any reason (in-
`cluding progressive disease, toxicity, or withdrawal of con-
`sent) and who therefore did not have two sequential PSA
`values for response assessment, were nevertheless consid-
`ered to be nonresponders. Progressive disease was defined
`by a PSA increase of ⱖ 50% above nadir, on at least two
`successive occasions at least 1 month apart, with a mini-
`mum rise of 5 ng/mL [17]. Disease progression was also
`defined by new lesions on bone or computed tomography
`scan, or for patients with pre-existing measurable disease, a
`greater than 25% increase in the sum of the products of the
`perpendicular diameters of all measurable lesions. Survival
`was measured from the time of randomization to the time
`of death, and time to PSA progression was measured per
`Consensus Criteria [17].
`
`Statistical Methods
`The primary objective of this trial was to compare the
`response rates (composite end point, as defined above) of
`AAWD alone versus AAWD combined with ketoconazole
`and hydrocortisone. However, secondary objectives were to
`compare the percent of patients with a posttherapy decline
`in PSA of ⱖ 50%, documented on at least two successive
`occasions, at least 4 weeks apart, per the PSA Consensus
`Criteria [17]. Secondary objectives were also to evaluate the
`posttherapy PSA decline in patients who received ketocon-
`azole and hydrocortisone after developing progressive dis-
`ease despite AAWD. The study also sought to correlate
`posttherapy decline in PSA of ⱖ 50% PSA with survival, as
`well as to evaluate the prognostic value of several pretreat-
`ment patient characteristics. Finally, the study evaluated the
`relationship of pretreatment adrenal androgen levels, and
`changes in adrenal androgen levels with response to therapy
`and survival.
`With 119 patients per arm, and a one-sided ␣of .05, the
`independent two-group binomial test had 80% power to
`
`www.jco.org
`
`1027
`
`Downloaded from jco.ascopubs.org on July 20, 2016. For personal use only. No other uses without permission.
`Copyright © 2004 American Society of Clinical Oncology. All rights reserved.
`
`

`

`Small et al
`
`detect a difference in the percentage of patients with a
`partial or complete (composite) response from 25% in the
`AAWD group, to 40% in the AAWD ⫹ ketoconazole arm.
`Allowing for a 5% ineligibility rate, the target sample size
`was 250 patients.
`The study was monitored by the CALGB Data Safety
`and Monitoring Board. Planned interim analysis used the
`Lan-DeMets analog of the O’Brien-Fleming sequential
`boundary to maintain the overall level of significance of .05
`[18]. The Lans-DeMets stopping rule was applied to the
`composite response end point. At the final analysis, the ␣
`level was .045. An intent-to-treat approach was used in the
`analysis. Pearson ␹2 and Fisher’s exact test were used to
`compare the two arms with regard to response rates (com-
`posite end point), objective response rates, and 50% decline
`in PSA [19]. Exact confidence intervals based on the bino-
`mial distribution were used to estimate 95% CIs for the
`response rates. The Kaplan-Meier product-limit method
`was used to estimate overall survival and PSA progression-
`free survival by the two arms [20], and the log-rank test
`was used to compare the two arms on these outcomes
`(overall survival and PSA progression-free survival). The
`proportional hazards model was used to assess important
`factors for predicting survival time adjusting on the
`stratification factors.
`The relationship between overall survival and 50% de-
`cline in PSA from baseline was explored. To minimize “lead
`time bias,” landmark analyses were performed at 4, 8, 12,
`and 16 weeks postrandomization [21]. This method selects
`a fixed time point after initiation of therapy as a “landmark”
`and excludes patients who died before reaching the land-
`mark (eg, 8 weeks). Further, the patients alive at the land-
`mark were classified as responders or nonresponders de-
`pending on their 50% decline in PSA before the landmark.
`In these analyses, survival duration was defined as the time
`between the landmark (eg, 8 weeks) and death. The rela-
`tionship between survival duration, and PSA decline was
`examined. For the primary end point, a one-sided ␣⫽ .05
`was used to compute the 95% CI and the P value. For the
`secondary objectives and analyses, tests were performed
`using a two-sided ␣⫽ .05.
`
`RESULTS
`
`Patient Characteristics
`Two hundred sixty patients were enrolled onto this
`study. No consistent approach to screening patients for this
`trial was mandated. One hundred thirty-two were random-
`ized to AAWD alone and 128 were randomized to AAWD
`and ketoconazole. Patient characteristics, including strati-
`fication variables, are summarized in Table 1. With regard
`to stratification variables, approximately 36% of patients
`(on both arms) received prior flutamide, 59% had prior
`bicalutamide, and 4% to 5% had received nilutamide. Ap-
`
`Table 1. Baseline Characteristics
`
`Age, years
`Median
`Interquartile range
`Race, % white
`Sites of disease, %ⴱ
`Bone metastases
`Measurable disease
`Lymph node involvement
`Lung metastases
`Liver metastases
`Lymph node metastases only
`Performance status (0 to 1), %
`Opioid analgesic use, %
`Hemoglobin, g/dL
`Median
`Interquartile range
`PSA, ng/mL1
`Median
`Interquartile range
`Alkaline phosphatase, U/L
`Median
`Interquartile range
`LDH U/L
`Median
`Interquartile range
`Creatinine, mg/dL
`Median
`Interquartile range
`Prior therapy, %
`Flutamide
`Bicalutamide
`Nilutamide
`Androgen deprivation, %
`Intermittent
`Continuous
`Initial therapy for advanced
`disease, %
`CAB
`Monotherapy/ later antiandrogen
`
`AAWD
`(n ⫽ 132)
`
`AAWD and
`Ketoconazole
`(n ⫽ 128)
`
`Total
`(N ⫽ 260)
`
`71
`66-76
`78
`
`72
`64-76
`81
`
`72
`65-76
`79
`
`86
`31
`28
`5
`5
`5
`93
`30
`
`84
`40
`34
`6
`10
`8
`93
`28
`
`84
`35
`31
`5
`6
`7
`93
`29
`
`12.6
`11.7-13.3
`
`12.6
`11.1-13.6
`
`12.6
`11.6-13.5
`
`58
`17-162
`
`125
`91-239
`
`58
`20-137
`
`120
`85-225
`
`58
`20-151
`
`124
`90-235
`
`200
`171-405
`
`215
`189-409
`
`210
`176-405
`
`1.0
`0.9-1.2
`
`1.1
`1.0-1.3
`
`1.1
`0.9-1.3
`
`35.6
`59.1
`5.3
`
`12.9
`87.2
`
`59.9
`40.1
`
`37.5
`58.6
`3.9
`
`16.4
`83.6
`
`59.4
`40.6
`
`36.5
`58.9
`4.6
`
`14.6
`85.4
`
`59.6
`40.4
`
`Abbreviations: AAWD, antiandrogen withdrawal; PSA prostate-specific
`antigen; LDH, lactate dehydrogenase; CAB, combined androgen blockade.
`ⴱPatients may have more than one metastasis.
`
`proximately 15% of patients had received prior intermittent
`androgen deprivation, and approximately 60% had re-
`ceived initial combined androgen blockade (luteinizing
`hormone-releasing hormone analog plus antiandrogen).
`Imaging studies were positive for metastatic disease in 97%
`and 94% of patients in the AAWD and AAWD and keto-
`conazole arms, respectively. The two arms were similar
`regarding age, sites of disease, requirement for opioid anal-
`gesics, and a variety of pretreatment prognostic factors,
`including performance status, hemoglobin, PSA, alkaline
`phosphatase, LDH, and creatinine. The median age of pa-
`tients in both arms was just older than 70 years, and 93%
`
`1028
`
`JOURNAL OF CLINICAL ONCOLOGY
`
`Downloaded from jco.ascopubs.org on July 20, 2016. For personal use only. No other uses without permission.
`Copyright © 2004 American Society of Clinical Oncology. All rights reserved.
`
`

`

`AAWD and Ketoconazole for Prostate Cancer
`
`Table 2. Summary of Clinical Outcome
`
`PSA decline ⱖ 50%
`No. of patients
`%
`95% CI
`Objective response rate
`No. of patients
`%
`95% CI
`Survival time, months
`Median
`95% CI
`Time to PSA progression in
`PSA responders,
`months
`Median
`95% CI
`No. of patients
`
`AAWD Alone
`(n ⫽ 132)
`
`AAWD and
`Ketoconazole
`(n ⫽ 128)
`
`34/128
`15/132
`27
`11
`7% to 17% 20% to 35%
`
`10/50
`1/41
`20%
`2%
`0.1% to 11% 11% to 32%
`
`P
`
`.002
`
`.020
`
`16.7
`14.3 to 21.5
`
`15.3
`13.40 to 19.5
`
`.936
`
`5.9
`5.3 to 10.1
`15
`
`8.6
`5.7 to 20.4
`34
`
`.063
`
`Abbreviations: AAWD, antiandrogen withdrawal; PSA, prostate-
`specific antigen.
`
`had a performance status of 0 or 1. The median serum PSA
`levels at study entry was 58 ng/mL. Thirty-one percent of
`patients in the AAWD arm and 39% of patients in the
`AAWD and ketoconazole arms had measurable disease.
`More than 80% in both arms had bone metastases, and
`approximately one-third had lymph node involvement.
`Approximately 30% of patients were using opioid analge-
`sics at the time of study entry.
`Clinical Outcome
`Relevant clinical outcomes, including PSA decline, ob-
`jective response rate, overall survival, and time to PSA pro-
`gression are summarized in Table 2. Overall, 15 (11%; 95%
`CI, 7% to 17%) of 132 patients undergoing antiandrogen
`withdrawal alone experienced a ⱖ 50% decline in PSA. By
`contrast, 34 (27%; 95% CI, 20% to 35%) of 128 patients
`who underwent AAWD and received simultaneous keto-
`conazole had a ⱖ 50% decline in PSA (P ⫽ .002). In patients
`with a ⱖ 50% decline in PSA, the subsequent median time
`to PSA progression was 5.9 months (95% CI, 5.3 to 10.1
`months) and 8.6 months (95% CI, 5.7 to 20.4 months) in
`the AAWD alone and AAWD and ketoconazole arms, re-
`spectively (log-rank P ⫽ .063). Figure 1 demonstrates the
`overall PSA progression-free survival by treatment arm for
`those patients who had a 50% decline in PSA. Objective
`responses in measurable disease were observed in 1 (2%;
`95% CI, 0.13% to 11%) of 41 of patients treated with
`AAWD alone, compared with 10 (20%; 95% CI, 11% to
`32%) of 50 in the AAWD and ketoconazole arm (P ⫽ .02).
`While no longer currently in use, when the composite end
`points described were applied, 8 (6%) of 132 of patients
`treated with AAWD had a response, compared with 22
`
`Fig 1. Overall prostate-specific antigen (PSA) progression-free survival by
`treatment arm in patients with 50% decline in PSA. AAWD, antiandrogen
`withdrawal.
`
`(17%) of 128 of patients on the AAWD/ketoconazole arm
`(one-sided P ⫽ .004).
`Eighty-two percent of patients (108 of 132) treated
`with AAWD alone ultimately received “deferred” ketocon-
`azole/hydrocortisone therapy. Twenty-four patients as-
`signed to initial AAWD alone never received deferred keto-
`conazole for a variety of reasons,
`including disease
`progression and withdrawal of consent. Of the 108 patients
`receiving deferred ketoconazole, a decline in PSA of ⱖ 50%
`was observed in 35 (32%) of 108 patients. An objective
`response was seen in 3 (7%) of 41 patients treated with
`AAWD followed by deferred ketoconazole. Although the
`study was not designed to compare the aggregate PSA de-
`clines and objective responses in patients treated with si-
`multaneous versus sequential AAWD and ketoconazole,
`there did not seem to be an advantage of either approach
`over the other. In aggregate, 40 (30%) of 132 patients
`treated with AAWD followed by ketoconazole had a PSA
`decline of ⱖ 50% compared with 34 (27%) of 128 of pa-
`tients treated with simultaneous AAWD and ketoconazole.
`The median survival time was 16.7 months (95% CI, 14.3 to
`21.5 months) in the AAWD-alone arm, and 15.3 months
`(95% CI, 13.4 to 19.5 months) in the group of patients who
`received simultaneous AAWD and ketoconazole therapy
`(two-sided P ⫽ .936; Fig 2).
`For those patients receiving sequential AAWD fol-
`lowed by ketoconazole, prior PSA “response” (decline of ⱖ
`50%) after AAWD seemed to be associated with a higher
`likelihood of PSA response to subsequent ketoconazole.
`Overall, 35 patients had a PSA response when treated with
`deferred ketoconazole. Ten of these PSA responses oc-
`curred in the group of 15 patients who had obtained a prior
`PSA response to AAWD (10 of 15; 67%), whereas an addi-
`tional 25 patients had a response to deferred ketoconazole
`from among a group of 117 patients (25 of 117; 21%) who
`had failed to achieve a PSA response after AAWD.
`
`www.jco.org
`
`1029
`
`Downloaded from jco.ascopubs.org on July 20, 2016. For personal use only. No other uses without permission.
`Copyright © 2004 American Society of Clinical Oncology. All rights reserved.
`
`

`

`Small et al
`
`Fig 2. Overall survival by treatment arm. AAWD, antiandrogen
`withdrawal.
`
`A proportional hazards model was used to identify
`predictors of overall survival. Pretreatment PSA, alkaline
`phosphatase, LDH, and hemoglobin levels, each dichoto-
`mized at the median; the presence or absence of weight loss
`at the time of study entry; and prior therapy with flutamide
`but not treatment arm, were predictors of survival (Table
`3). In addition, using a 4-week landmark analysis, a ⱖ 50%
`decline in PSA was a strong predictor of survival— 41
`months in patients with a ⱖ 50% PSA decline versus 13
`months in patients without a PSA decline (P ⬍ .0001; Fig 3).
`Toxicity
`Overall, 7% of patients undergoing AAWD alone had a
`grade 3 or 4 toxicity (all presumably attributed to causes
`
`Table 3. Predictors of Survival Time in Proportional Hazards Model
`(all patients)
`
`Variable
`
`Weight loss at baseline, ⬎ 5%
`(yes or no)
`Pretreatment PSA, ⬎ 58 v
`58 mg/mL
`Alkaline phosphatase, ⬎ 124 v
`124 U/L
`LDH, ⬎ 210 v 210
`Performance status, 2 v 0 or 1
`Hemoglobin, ⬎ 12.6 v 12.6
`Prior therapy, flutamide v
`nilutamide or bicalutamide
`Androgen deprivation, continuous v
`intermittent
`Initial therapy for advanced disease,
`monothrapy v CAB ⫹
`antiandrogen
`Treatment arm, AAWD ⫹
`ketoconazole v AAWD
`
`Hazard
`Ratio
`
`95% CI
`
`P
`
`2.63
`
`1.48 to 4.65 ⬍ .001
`
`1.60
`
`1.17 to 2.19
`
`1.50
`
`1.10 to 2.03
`
`1.33
`1.30
`0.64
`1.53
`
`1.00 to 1.78
`0.70 to 2.44
`0.47 to 0.86
`1.12 to 2.08
`
`1.29
`
`0.88 to 1.93
`
`1.18
`
`0.88 to 1.57
`
`.003
`
`.010
`
`.051
`.402
`.004
`.008
`
`.208
`
`.273
`
`1.09
`
`0.82 to 1.43
`
`.555
`
`Abbreviations: PSA, prostate specific antigen; LDH, lactate dehydroge-
`nase; CAB, combined androgen blockade; AAWD, antiandrogen with-
`drawal.
`
`Fig 3. Kaplan-Meier plot of overall survival as a function of ⱖ 50% decline
`in prostate-specific antigen (PSA), using a 1-month landmark analysis.
`
`other than AAWD itself), but no one toxicity occurred in
`more than 3% of patients. Twenty-one percent of patients
`receiving ketoconazole had grade 3 and 4 toxicities. The
`most common toxicities with ketoconazole were neurologic
`toxicity (4%), which consisted of motor neuropathy and
`ototoxicity, and malaise or fatigue (3%; Table 4). Grade 3 or
`4 hepatic toxicity was observed in 2% of patients receiving
`ketoconazole, and was not more common than in the
`AAWD arm.
`Adrenal Androgen Levels
`Baseline plasma adrenal androgen levels are available
`on 213 patients, 113 in the AAWD-alone arm, and 100 in
`the AAWD and ketoconazole arm. One hundred sixty-four
`patients had levels drawn after 1 month of therapy; 81, after
`3 months; and 111, at progression, which occurred at a
`median of 3.1 months after starting therapy (95% CI, 2.1 to
`4.9 months). Baseline median DHEAS and androstendione
`levels were within the normal range, while baseline DHEA
`levels were slightly elevated (Table 5). Baseline median
`DHEA, DHEAS, and androstendione levels were similar in
`both arms (data not shown). The median baseline testoster-
`
`Table 4. Treatment-Related Grade 3 and 4 Toxicities
`
`% of Patients
`
`AAWD Alone
`(n ⫽ 124)
`
`AAWD and
`Ketoconazole
`(n ⫽ 124)
`
`Hepatic toxicity
`Anorexia
`Neurotoxicity
`Cardiotoxicity
`Pulmonary
`Coagulation
`Nausea/vomiting
`Malaise/fatigue
`
`4
`0
`0
`0
`0
`0
`0
`0
`
`Abbreviation: AAWD, antiandrogen withdrawal.
`
`2
`2
`4
`1
`2
`1
`1
`4
`
`1030
`
`JOURNAL OF CLINICAL ONCOLOGY
`
`Downloaded from jco.ascopubs.org on July 20, 2016. For personal use only. No other uses without permission.
`Copyright © 2004 American Society of Clinical Oncology. All rights reserved.
`
`

`

`AAWD and Ketoconazole for Prostate Cancer
`
`Table 5. Median Adrenal Androgen Levels in AAWD Plus Ketoconazole-Treated Patients
`
`Baseline (n ⫽ 100)
`
`Month 1 (n ⫽ 73)
`
`At Progression (n ⫽ 56)
`
`Adrenal Androgen
`
`Normal Range
`
`Median
`
`25th-75th Percentile
`
`Median
`
`25th-75th Percentile
`
`Median
`
`25th-75th Percentile
`
`DHEA
`DHEAS
`
`Androstendione
`
`(0.2-1.5 ng/mL)
`(31-4,668 ng/
`mL)
`(0.3-3.1 ng/mL)
`
`2.1
`317
`
`0.6
`
`1.4-3.3
`144-696
`
`0.5-1.1
`
`1.0
`30
`
`0.30
`
`0.8-1.4
`1-70
`
`0.2-0.4
`
`1.3
`116ⴱ
`
`0.45ⴱ
`
`0.8-2.0
`18-270
`
`0.3-0.6
`
`Abbreviations: AAWD, antiandrogen withdrawal; DHEA, dehydroeplandrosterone; DHEAS, DHEA sulfate.
`ⴱDifference between levels at month 1 and levels at progression are significant at P ⫽ .0001.
`
`one level was 13 ng/mL, and did not change over time in
`either treatment arm. DHEA, DHEAS, and androstendione
`levels did not change appreciably over time in the AAWD
`group (data not shown). By contrast, in the AAWD and
`ketoconazole group, there was a decline in levels of all three
`adrenal androgens (DHEA, DHEAS, androstendione), ac-
`counting for declines from baseline of 54%, 90%, and 58%,
`respectively, at 1 month. There was a rise in all three adrenal
`androgen levels at the time of disease progression, though
`not back to baseline (Table 5). The difference between
`DHEAS and androstendione levels at month 1 and at the
`time of progressive disease are significant (P ⫽ .0001).
`
`DISCUSSION
`
`This prospective, randomized phase III trial has compared
`changes in PSA levels as well as objective responses in met-
`astatic AiPCa patients treated with either AAWD alone or
`with AAWD plus simultaneous ketoconazole and hydrocorti-
`sone replacement therapy. No difference in survival was ob-
`served, though the planned use of deferred ketoconazole in the
`AAWD-alone arm may have blunted a treatment effect.
`This study has demonstrated that PSA declines and
`objective responses after AAWD are uncommon. A PSA
`decline of ⱖ 50% was observed in 11% of patients under-
`going AAWD, while an objective response proportion of
`2% was noted. The median time to PSA progression was
`brief, at 5.9 months. The AAWD phenomenon has been
`appreciated for many years [6-8], though there are few
`prospective trials carefully assessing its frequency. To our
`knowledge, this is the single largest prospective study of
`AAWD. These data indicat