5403.Full-Text.Pdf

Vol. 10, 5403–5417, August 15, 2004 Clinical Cancer Research 5403

Breast Cancer Chemoprevention Phase I Evaluation of Biomarker Modulation by Arzoxifene, a Third Generation Selective Estrogen Receptor Modulator

Carol J. Fabian,1 Bruce F. Kimler,1 various hormones and growth factors. Tissue from biopsies Julie Anderson,1 Ossama W. Tawfik,1 (estrogen receptor ؉ and/or progesterone receptor ؉) and Matthew S. Mayo,1 William E. Burak, Jr.,2 re-excision specimens was evaluated immunohistochemi- 3 4 cally for proliferation (Ki-67 by MIB-1 and proliferating cell Joyce A. O’Shaughnessy, Kathy S. Albain, nuclear antigen) and other biomarkers. 5 6 David M. Hyams, G. Thomas Budd, Results: In both trials, increases in serum sex hormone Patricia A. Ganz,7 Edward R. Sauter,8 binding globulin were noted, as were decreases in insulin- Samuel W. Beenken,9 William E. Grizzle,9 like growth factor (IGF)-I and the IGF-I:IGF binding pro- John P. Fruehauf,10 Dora W. Arneson,11 tein-3 ratio (P < 0.007 versus control/placebo). For 45 evalu- 12 13 able women in Phase IA, decreases in proliferation indices James W. Bacus, Michael D. Lagios, were more prevalent for arzoxifene (particularly 20 mg) 14 14 Karen A. Johnson, and Doris Browne than for controls. For 58 evaluable women in Phase IB, a 1University of Kansas Medical Center, Kansas City, Kansas; 2Ohio decrease in estrogen receptor expression for arzoxifene was 3 ؍ State University, Columbus, Ohio; US Oncology, Inc., Dallas, Texas; observed compared with no change with placebo (P 4Loyola University Medical Center, Maywood, Illinois; 5Desert Comprehensive Cancer Center, Palm Springs, California; 6Cleveland 0.0068). However, decreases in proliferation indices for ar- Clinic Foundation, Cleveland, Ohio; 7University of California Los zoxifene were not statistically different from placebo, per- Angeles, Los Angeles, California; 8University of Missouri-Columbia, haps due to a confounding effect of stopping hormone re- Columbia, Missouri; 9University of Alabama-Birmingham, 10 placement therapy before entry. Birmingham, Alabama; Oncotech, Inc., Irvine, California; Conclusion: Given the favorable side effect profile and 11Midwest Research Institute, Kansas City, Missouri; 12Bacus Laboratories, Inc., Elmhurst, Illinois; 13St. Mary’s Hospital, San the biomarker modulations reported here, arzoxifene re- Francisco, California; and 14Division of Cancer Prevention, National mains a reasonable candidate for additional study as a Cancer Institute, Bethesda, Maryland breast cancer chemoprevention agent.

INTRODUCTION ABSTRACT Arzoxifene (LY353381) is a third-generation selective es- Purpose: Arzoxifene, a new selective estrogen receptor trogen receptor modulator similar to raloxifene but modified to modulator with strong breast antiestrogen activity and ab- improve bioavailability as well as estrogen antagonist potency sence of uterine agonist activity, was explored as a potential in the breast and uterus (1). In vitro and xenograft studies have chemoprevention agent. We performed a multi-institutional shown arzoxifene to be more potent than tamoxifen, 4-OH evaluation of arzoxifene in women with newly diagnosed tamoxifen, or raloxifene in inhibiting the growth of tamoxifen- ductal carcinoma in situ or T1/T2 invasive cancer. sensitive, estradiol-stimulated MCF-7 cells (2). Arzoxifene also Experimental Design: In a Phase IA trial, 50 pre- or was able to suppress the growth of T47D tamoxifen-resistant postmenopausal women were randomized to 10, 20, or 50 tumors but not MCF-7 tamoxifen-resistant tumors in xenograft mg of arzoxifene daily in the interval between biopsy and models (3). The two tamoxifen-resistant models differ in that the re-excision or were enrolled as no-treatment controls. In a resistant MCF-7 cells have reduced levels of the estrogen re- Phase IB trial, 76 postmenopausal women were randomized ceptor corepressor NCOR, whereas resistant T47D cells have to 20 mg of arzoxifene versus matched placebo. Serum spec- increased levels of estrogen receptor coactivators such as AIB1 imens collected at entry and at re-excision were assayed for (3–5). Arzoxifene was found to be more effective than equimo- lar doses of raloxifene and equivalent to tamoxifen in the nitrosomethylurea rat breast cancer prevention model (2). Pre- clinical studies in ovariectomized rats indicate favorable effects Received 1/28/04; revised 5/7/04; accepted 5/17/04. on cholesterol, bone mineral density, and uterine weight, which Grant support: NO1-CN-85035 from the Chemoprevention Branch, support the potential utility of arzoxifene as a prevention agent. Cancer Prevention Research Program-Cancer Control, Division of Can- Indeed, in preclinical studies arzoxifene was more potent than cer Prevention of the National Cancer Institute, NIH. Notes: C. Fabian and B. Kimler contributed equally to this work. raloxifene in the prevention of bone loss (1, 6). Low-dose The costs of publication of this article were defrayed in part by the arzoxifene and a novel rexinoid have been demonstrated to be payment of page charges. This article must therefore be hereby marked synergistic in the rat model of breast cancer (7). advertisement in accordance with 18 U.S.C. Section 1734 solely to Animal models predicted antitumor activity at equivalent indicate this fact. Requests for reprints: Carol J. Fabian, University of Kansas Medical human doses of 10 mg/day (2). Studies in healthy postmeno- Center, 3901 Rainbow Boulevard, Kansas City, KS 66160. Phone: pausal volunteers and in women with metastatic disease indi- (913) 588-7791; Fax: (913) 588-3679; E-mail: [email protected]. cated that the pharmacokinetics of arzoxifene were linear over a

wide dose range (8, 9). Pharmacodynamic changes were ob- willing to participate served as nonrandomized no-treatment served starting at the 10 mg per day dose and included reduced controls. Women on hormone replacement therapy discontinued levels of low-density lipoprotein cholesterol, serum fibrinogen, this medication before study entry. Doses selected for the Phase antithrombin III activity, follicle-stimulating hormone (FSH), IA trial were 10, 20, and 50 mg/day (10-mg and 50-mg tablets luteinizing hormone (LH), and the bone turnover biomarkers supplied by Eli Lilly via the National Cancer Institute). Dose osteocalcin and urine type I collagen fragments, whereas an selection was based on the favorable pharmacodynamic effects increase was observed for sex hormone binding globulin (9). of arzoxifene in doses as low as 10 mg/day (9) and reported Significant antitumor activity has been reported for tamox- responses at 20 and 50 mg/day in advanced breast cancer. ifen-naive or -sensitive subjects in Phase II trials of women with Arzoxifene has a terminal elimination half-life of 30–35 h (9). metastatic disease. Baselga et al. (10) reported a complete and On the basis of the pharmacokinetic profile of arzoxifene, steady partial response rate of 43% for arzoxifene at 20 mg/day and state was not predicted to be reliably reached during a 2–4 week 27% for arzoxifene at 50 mg/day in a cohort with a median age trial using once daily dosing; therefore, a loading dose consist- of 70 and in whom only 9% had received prior tamoxifen. Only ing of twice the assigned daily dose was used on day 1 and day 10% of tamoxifen refractory subjects responded to 20–50 mg 2 for both the Phase IA and Phase IB trials. Subjects were daily arzoxifene (11). No uterine agonist effects were reported instructed to take their study drug with a meal. Once a subject in women who had not been exposed to tamoxifen previously had been randomly assigned a drug dose, the dose a subject was (11). receiving was known to investigators. Despite preclinical and clinical evidence of partial cross- The Phase IB trial as originally activated was a double- resistance with tamoxifen, the demonstrated efficacy of arzox- blind, three-way, 1:1:1 randomization among arzoxifene 20 ifene in advanced breast cancer, lack of uterine agonist activity, mg/day, tamoxifen 20 mg/day, and matching placebos for both and favorable pharmacodynamic effects on bone and lipids drugs (supplied by McKesson Laboratories under contract to the make it an attractive selective estrogen receptor modulator for National Cancer Institute). The trial was restricted to postmeno- prevention. pausal women who had not received hormone replacement A number of studies have demonstrated that selective es- therapy within 30 days of biopsy. However, accrual was ex- trogen receptor modulators likely to be effective in breast cancer tremely slow (14 subjects in 8 months) due in part to the treatment and prevention will show a reduction in proliferation exclusion of women who had received hormone replacement in estrogen receptor-positive tumors after 2–4 weeks of treat- therapy within 30 days of biopsy and the difficulty of explaining ment (12–16). a three-way randomization to potential subjects. Consequently, Therefore, before beginning large-scale prevention studies the protocol was amended to allow women who had received with arzoxifene, a short-term Phase IA/IB preoperative biomar- hormone replacement therapy in the peribiopsy period to enter ker modulation study in women with newly diagnosed breast the study providing hormone replacement therapy was discon- cancer was initiated by a multi-institutional group as part of tinued before study entry (subjects were stratified according to a Division of Cancer Prevention, National Cancer Institute- hormone replacement therapy use within 30 days of biopsy). sponsored contract. In the Phase IA portion of the trial, women The protocol was also changed to a design of two sequential with core biopsy evidence of ductal carcinoma in situ and/or phases: 60 subjects in a 2:1 randomization between arzoxifene invasive cancer were randomized to one of several doses of and matching placebo to be followed by 60 subjects in a 2:1 arzoxifene in the interval between core biopsy and re-excision. randomization between tamoxifen and matching placebo. Sub- The purpose of the Phase IA trial was to select the lowest dose sequently, a decision was made to omit the tamoxifen phase and associated with a reduction in proliferation as well as identify to instead continue subject accrual with a target accrual of ϳ60 other modulated biomarkers consistent with the presumed subjects on the arzoxifene arm. The blinded assignment between mechanism of action (17, 18). In the Phase IB trial, the dose arzoxifene and placebo was maintained throughout the study selected from the Phase IA trial was compared in a randomized and subsequent analysis. Only the biostatistician at University of double-blind fashion to placebo with the primary end point of Kansas Medical Center not otherwise associated with this trial reduction in proliferation (17). who prepared the randomization assignments and the investiga- We report the results of Phase IA and Phase IB biomarker tional pharmacists at each participating institution were aware of modulation trials of arzoxifene administered to women with the study agent assignments. For analysis, the study agent as- hormone receptor-positive ductal carcinoma in situ and/or signments were revealed to the University of Kansas Medical T1-T2 invasive breast cancer in the interval between diagnostic Center biostatisticians conducting the analyses. biopsy and definitive surgical treatment. Table 1 summarizes study design, eligibility, and biomarker endpoints for the Phase IA and Phase IB trials. Subject Selection/Eligibility. Eligible subjects for both MATERIALS AND METHODS Phase IA and IB trials were women with a clinical ductal Study Design. In the Phase IA trial, eligible pre- and carcinoma in situ, T1, or T2/N0 tumor, who had recently un- postmenopausal women were randomized to one of three doses dergone a core or excisional biopsy in which tumor was present of arzoxifene administered daily in the planned interval between at the margins and who were anticipating a minimum of 2 weeks biopsy and re-excision. The planned interval was not altered by between biopsy and re-excision. Premenopausal women eligible participation in the trial, but subjects were not eligible if the for the Phase IA trial must have had a negative pregnancy test interval was anticipated to be Ͻ2 weeks or Ͼ6 weeks. Women and must have been willing to use a barrier type of contraception not wishing to receive drug but who were otherwise eligible and during trial participation. When hormone replacement therapy

Table 1 Study design, eligibility, and biomarker endpoints for Phase IA and Phase IB trials Phase IA Phase IB Study design Randomized to arzoxifene 10, 20, or 50 mg/day Randomized to arzoxifene 20 mg/day versus placebo (2:1 ratio) Nonrandomized no-treatment controls 2–6 weeks of study drug (median, 2 weeks) 2–6 weeks of study drug (median, 2 weeks) Eligible cohort DCIS/T1-T2 invasive cancers DCIS/T1-T2 invasive cancers Pre- and postmenopausal women Postmenopausal women HRT allowed before study entry HRT allowed before study entry Likely to be ERϩ or PRϩ* Likely to be ERϩ or PRϩ* Main endpoint Proliferation (Ki-67/PCNA) Proliferation (Ki-67/PCNA) Secondary tissue endpoints ER, PR, EGFR, Bcl-2, Her-2/neu, p53 ER, PR, EGFR, Bcl-2, Her-2/neu, p53, pERK, IGFR, TNF-␣, MVD Secondary serum endpoints IGF-I, IGFBP-3, estradiol, progesterone, estrone, IGF-I, IGFBP-3, estradiol, estrone, FSH, LH, SHBG FSH, LH, SHBG, TBG Abbreviations: IGFBP, IGF binding protein; DCIS, ductal carcinoma in situ; SHBG, sex hormone binding globulin; TGB, thyroxine-binding globulin; ER, estrogen receptor; PR, progesterone receptor; pERK, phosphorylated extracellular regulated kinase; EGFR, epidermal growth factor receptor; IGFR, IGF receptor; TNF, tumor necrosis factor; MVD, microvessel density; HRT, hormone replacement therapy. * Subjects were eligible if ER/PR status was known to be positive or if unknown and the tumor was low-intermediate grade (likely to be ERϩ and/or PRϩ). However, only those subjects with biopsy specimen later confirmed to be ERϩ and/or PRϩ were included in the analysis for biomarker response.

(or oral contraceptive) use before diagnosis was allowed, it was thyroxine-binding globulin). For the Phase IB trial, serum as- required that they be discontinued before study entry. To be says were conducted at the Center for Reproductive Endocri- eligible, the biopsy must have been fixed in 10% neutral- nology Laboratories, University of Kansas Medical Center (Dr. buffered formalin according to a prescribed protocol and the Paul Terranova) using enzyme immunoassays from Diagnostic tumor judged as either non-high grade according to the institu- Systems Laboratories, Inc. (Webster, TX; estradiol, estrone, tional pathologist and/or found to be estrogen receptorϩ or FSH, LH, IGF-I, and IGFBP-3) and Diagnostic Biochemicals progesterone receptorϩ by institutional standards with enough (London, Ontario, Canada; sex hormone binding globulin). tumor remaining in the tissue block for 24–36 additional sec- Tissue Assays. The tissue biomarkers assessed in Phase tions. Women were required to be in generally good health with IA and IB are listed in Table 2. In addition to the primary study no prior history of deep venous thrombosis, stroke, or uterine end points, a number of other tissue biomarkers were assessed cancer. Normal organ function was required with Ͼ10 g/dl that might be predictive and/or reflective of successful selective hemoglobin, Ͼ1000/␮l absolute granulocyte count, Ͻ1.5 mg/dl estrogen receptor modulator treatment, including predictive creatinine, Ͼ3.0 g/dl albumin, Ͻ1.5 mg/dl bilirubin, Ͻ100 markers such as estrogen receptor, progesterone receptor, and units/l aspartate aminotransferase, and Ͻ200 units/l alkaline Her-2/neu (22, 23); markers of angiogenesis such as throm- phosphatase. bospondin, CD31, and p53 incorporated into an angiogenesis Serum Biomarkers. An average of 15 ml of serum was index, vascular endothelial growth factor, and microvessel den- obtained before entry and again at the off-study point for estra- sity (24–26); and growth and survival signaling markers such as diol, estrone, FSH, LH, insulin-like growth factor (IGF-I), bcl-2, phosphorylated extracellular regulated kinase, tumor ne- insulin-like growth factor binding protein type 3 (IGFBP-3), crosis factor ␣, and insulin-like growth factor receptor (27–30) progesterone, sex hormone binding globulin, and thyroxine- Tissue was fixed in 10% neutral-buffered formalin at the indi- binding globulin for the Phase IA trial. The same assays were vidual sites for Ն6 h but Յ24 h and then processed to paraffin performed for the Phase IB trial, with the exception that pro- blocks. After sufficient tissue had been used at the site for gesterone and thyroxine-binding globulin were omitted. Serum diagnostic purposes, blocks were sent to University of Kansas was stored at Ϫ70 to Ϫ80°C until analysis, with pre- and Medical Center for central sectioning, staining and interpreta- poststudy specimens assayed in the same run. Selective estrogen tion at the University of Kansas Medical Center Pathology receptor modulators such as tamoxifen had been described pre- Research Laboratories by O. W. T. Institutions were allowed to viously as modulating the risk biomarkers IGF-I, IGFBP-3, and send unstained tissue sections on slides provided these were sex hormone binding globulin (19–21). Modulation of thyroxine- processed and mailed such that slides could be stained within binding globulin and sex hormone binding globulin were also 72 h of sectioning of the block at the central laboratory at considered potential markers of selective estrogen receptor University of Kansas Medical Center. Staining and assessment modulator biological activity without regard to efficacy. Hor- of pre- and poststudy tissue assays were performed together. mone assays were performed to determine whether the systemic Sectioning of the block for all of the assays was performed at the hormone levels were substantially altered between the study same time. Automated stainers (Dako, Ventana) were used for entry time point and the re-excision time point. For the Phase IA staining all of the specimens, and known positive and negative trial, serum assays were conducted at Midwest Research Insti- controls were included with each run. Assessments were done tute (Kansas City, MO) using radioimmunoassays (for estradiol, with the CAS 200 image analyzer for Ki-67 (MIB-1 antibody), estrone, and progesterone) or immunoradiometric assays (for proliferating cell nuclear antigen (PCNA), estrogen receptor, FSH, IGF-1, IGFBP-3, sex hormone binding globulin, and progesterone receptor, and p53 and produced values for the

Table 2 Biomarkers employed in Phase IA and Phase IB trials Phase IA Phase IB Ki-67 (Immunotech MIB-1) Ki-67 (Dako MIB-1) PCNA (Dako PC10) PCNA (Dako PC10) ER (Ventana 6F-11 clone) ER (Biocare ER07) PR (Ventana 1A6 clone) PR (Dako PgR1294) EGFR (Zymed 3iG7) EGFR (Zymed 3iG7) Bcl-2 (Dako 124 clone) Bcl-2 (Dako 124 clone) Her-2/neu (Dako Herceptest) Her-2/neu (Dako Herceptest) p53 (Ventana BP53–11) p53 (Innogenex) Nuclear morphometry (Feulgens/BLISS) MVD (Dako factor VIII related antigen) Angiogenesis index (Oncotech–p53, TPS-1, CD-31) pERK (Cell Signaling Technology) IGF-1R␤ (Santa Cruz Biotechnology C20) TNF-␣ (obtained from Paul Terranova, KUMC) Abbreviations: ER, estrogen receptor; PR, progesterone receptor; pERK, phosphorylated extracellular regulated kinase; EGFR, epidermal growth factor receptor; TNF, tumor necrosis factor; MVD, microvessel density; KUMC, University of Kansas Medical Center.

percentage of cells staining positively. Manual assessment was able and change in the primary end point (proliferation) could be used for other biomarkers using an immunochemistry weighted evaluated. With this number of subjects, if a reduction in pro- index score (range, 0–4) developed by our immunochemistry liferation (Ki-67 or PCNA) was seen in at least 8 subjects, we consultant (W. E. G.; Ref. 31). Before beginning the study, all would be 89% confident that the probability of favorable mod- of the immunochemistry procedures were validated and ap- ulation is at least 50%. proved. The antibodies used are listed in Table 2. For the Phase The accrual goal for the amended Phase IB trial was 80 IA trial, assessment of nuclear morphometry was also performed subjects from which 60 subjects would be evaluable for the by O. W. T, B. F. K., and J. W. B. using an automated image primary end point biomarkers. This accrual goal was based on analysis system (BLISS; Bacus Laboratories, Inc., Elmhurst, the number of subjects needed to detect a large effect size of 0.8 IL). Fuelgen-stained specimens were scanned, areas of interest SDs in change in the proportion of cells expressing a prolifer- marked, and a z-score computed that incorporated a number of ation maker between the control and the treatment groups with parameters and represents their deviation from normalcy (32). an 80% power and type I error rate of 5%. From the Phase IA For the Phase IA trial, a subset of subjects was evaluated by results, this would project to an absolute decrease from 10% at J. P. F. for change in angiogenesis index (24). baseline to 4% or 5% at re-excision in the proportion of cells For the Phase IB trial, nuclear morphometric assessment expressing Ki-67 or PCNA, respectively. was omitted because it was noninformative in the Phase IA trial. Data Capture and Entry. Clinical data were entered The angiogenesis index was omitted because it was cumber- onto case report forms at individual sites, and audited data were some and required multiple slides, and the microvessel density entered into a computerized database at CCS Associates (Moun- was substituted. Several new assays were added (phosphoryl- tain View, CA). The clinical data were down loaded electroni- ated extracellular regulated kinase, tumor necrosis factor ␣, and cally into a joint biomarker and clinical database housed in the insulin-like growth factor 1 receptor), which might be inform- Biostatistical Unit of the University of Kansas Medical Center. ative regarding selective estrogen receptor modulator resistance Tissue biomarker and serum biomarker data were downloaded (33–37). PCNA, Ki-67, estrogen receptor, progesterone recep- electronically into this database as well. tor, p53, and Her-2/neu were analyzed with validated programs Statistical Analyses. For the Phase IA and Phase IB on an Automated Cellular Imaging System (ChromaVision trials, categorical variables were summarized by frequencies and Medical Systems, Inc., San Juan Capistrano, CA) image ana- percentages, and quantitative variables were summarized by lyzer and were verified manually. However, antibodies with medians and ranges. For the Phase IA trial, quantitative vari- predominately cytoplasmic and/or membrane localization (epi- ables were compared among the four groups using the Kruskal- dermal growth factor receptor, Bcl-2, insulin-like growth factor Wallis test. The Wilcoxon rank-sum test was used to perform receptor, phosphorylated extracellular regulated kinase, and tu- pair-wise comparisons on quantitative variables that were glo- mor necrosis factor ␣) were assessed manually by consensus bally different among the four groups. Fisher’s exact test was scoring (weighted index score) on a 0–4 scale. Microvessel used to compare categorical variables among the four groups. density (by factor VIII antibody) evaluation was performed by For the Phase IB trial, except for the evaluation of the primary assessing the areas of tumor microvessel density in a represent- endpoints (modulation of proliferation index and frequency of ative section. Vessels were counted in five random high- adverse events), all of the analyses were considered as explor- powered fields in the selected area and the mean vessel count atory. Thus, no corrections for multiple comparisons were made. from the five fields used. At least 1000 epithelial cells were Quantitative variables were compared between the two groups analyzed for each marker. using the Wilcoxon rank-sum test. The two-sample t test was Accrual Goals. Accrual goals for the Phase IA trial were also performed as per protocol. Whereas the conclusions are 10 subjects in each of the three (10, 20, and 50 mg/day) identical for each procedure, given the skewness of the distri- arzoxifene arms, and the nonrandomized no-treatment control butions for many variables, the Wilcoxon rank-sum test was the group, for whom prestudy and poststudy specimens were avail- preferred method of analysis. Categorical variables were com-

Table 3 Demographic characteristics of 45 subjects evaluable on the Phase IA trial Control (N ϭ 8) Arzoxifene (N ϭ 37) Total (N ϭ 45) P value Age (yrs) 55 61 61 0.74* Median (range) (40, 82) (42, 81) (40, 82) Number non-high-grade tumors 7 (88%) 29 (78%) 36 (80%) 1.00† Number DCIS only (per review at KUMC) 2 (25%) 6 (16%) 8 (18%) 0.62† Number premenopausal 3 (38%) 4 (11%) 7 (16%) 0.094† Number on HRT within 30 days of biopsy 3 (38%) 15 (41%) 18 (40%) 1.0† Time (days) between biopsy and study entry 11 10 10 0.40* Median (range) (7, 38) (0, 22) (3, 38) Time (days) on study agent - 15 15 Median (range) (10, 42) (10, 42) Time (days) between biopsy and re-excision 29 26 27 0.76* Median (range) (16, 47) (15, 55) (15, 55) Time (days) between stopping HRT and biopsy‡ 26 7 7 0.025* Median (range) (2, 27) (-13, 14) (-13, 27) Abbreviations: DCIS, ductal carcinoma in situ; HRT, hormone replacement therapy; KUMC, University of Kansas Medical Center. * Wilcoxon rank-sum test. † ␹2 Fisher exact test. ‡ Negative value denotes stopped before biopsy; positive value denotes stopped after biopsy.

pared using Fisher’s exact test. A two-way analysis of variance Forty women with newly diagnosed ductal carcinoma in with interaction was modeled to assess the effect of treatment situ, T1, or T2 tumors were randomized to 10, 20, and 50 mg of and hormone replacement therapy use within 30 days of biopsy arzoxifene daily (13, 13, and 14 subjects, respectively). In on the quantitative biomarkers. A significant interaction term addition, 10 subjects were registered as nonrandomized, no- would indicate a differing effect of treatment depending on treatment controls. Five subjects were not evaluable for tissue hormone replacement therapy status. based biomarkers, because there was no residual tumor available for analysis in the re-excision specimen (3 subjects) or because RESULTS the biopsy was both estrogen-receptor and progesterone-recep- Results of the Phase IA Trial tor negative (2 subjects) by immunohistochemistry performed at Accrual and Participating Institutions. From Novem- University of Kansas Medical Center. Thus, 45 Phase IA sub- ber 1998 to December 1999, 50 subjects were entered into the jects were evaluable for the primary biomarker endpoint includ- Phase IA trial after signing appropriate informed consents as ing 8 controls, 11 on the 10-mg arm, 13 on the 20-mg arm, and approved by the site Institutional Review Board. At least 315 13 on the 50-mg arm. Characteristics of these subjects are listed potential subjects were screened to accrue this number of sub- in Table 3. Median age was 61; 84% were postmenopausal. jects (the uncertainty resulted from variations in the definition Fifteen of 37 (41%) subjects on study drug and 3 of 8 (38%) and reporting of screening between sites). Accrual distribution no-treatment controls had received hormone replacement ther- included four from University of Kansas Medical Center, 15 apy within 30 days of biopsy. Eighty percent of biopsy speci- from Loyola University Medical Center, 14 from US Oncology, mens were grade I or II. Seventy-eight percent of women had 8 from Ohio State University, and 3 each from Cleveland Clinic invasive cancer in both their original biopsies and re-excision Foundation, University of California Los Angeles, and Desert specimens, whereas 13% had ductal carcinoma in situ in both Comprehensive Cancer Center. specimens. For 9% of subjects, the submitted biopsy and re-

Table 4 Adverse events* for 40 subjects receiving arzoxifene on the Phase IA trial Adverse event Grade 10 mg (N ϭ 13) 20 mg (N ϭ 13) 50 mg (N ϭ 14) Total (N ϭ 40) Menopausal symptoms† 1 0 2 2 4 (10%) 2 3 3 0 6 (15%) Headaches 1 1 0 1 2 (5%) 2 1 0 0 1 (2.5%) Nausea 1 1 3 3 7 (17.5%) 20000 Breast pain 1 2 1 1 4 (10%) 2 0 1 0 1 (2.5%) Other 1 2 5 5 12 (30%) 2 1 2 0 3 (7.5%) Overall adverse event’s worst grade 1 4 4 8 16 (40%) 2 3 4 0 7 (17.5%) * Interval between entry on study and re-excision surgery and considered to be drug-related. † Includes hot flashes and sweating.

Table 5 Baseline serum and tissue biomarker values (median and range) for 45 evaluable subjects on the Phase IA trial Arzoxifene Control 10 mg 20 mg 50 mg (N ϭ 8) (N ϭ 11) (N ϭ 13) (N ϭ 13) P value* Serum Estradiol (pg/ml) 52 (3, 636) 10 (3, 386) 15 (3, 189) 11 (3, 201) 0.64 Estrone (pg/ml) 41 (19, 828) 42 (25, 200) 31 (10, 189) 34 (10, 69) 0.11 FSH (mIU/ml) 53 (2, 72) 39 (14, 112) 47 (7, 131) 42 (10, 65) 0.86 LH (ng/ml) 35 (2, 88) 51 (9, 148) 47 (12, 122) 62 (22, 85) 0.48 IGF-I (ng/ml) 185 (117, 326) 171 (61, 252) 125 (62, 270) 111 (75, 252) 0.41 IGFBP-3 (␮g/ml) 3.6 (2.9, 5.2) 3.7 (2.8, 4.8) 3.7 (2.4, 4.8) 3.4 (2.5, 5.6) 0.52 IGF-1:IGFBP-3 (Molar ratio), ϫ 103 49 (33, 68) 42 (22, 55) 38 (25, 60) 42 (24, 58) 0.59 SHBG (nmol/l) 43 (20, 71) 40 (7, 78) 37 (13, 108) 41 (7, 75) 0.86 TBG (␮g/ml) 35 (27, 45) 30 (22, 47) 29 (21, 55) 30 (23, 53) 0.41 Progesterone (ng/ml) 0.3 (0.2, 2.4) 0.3 (0.1, 0.7) 0.2 (0, 0.8) 0.3 (0, 0.6) 0.58 Tissue Ki-67 (% positive) 9.5 (1, 41) 6.0 (1, 35) 10 (2, 40) 4.0 (1, 21) 0.41 PCNA (% positive) 1.0 (0.1, 15) 6.0 (0.1, 40) 9.0 (3, 61) 3.0 (1, 16) 0.012 ER (% positive) 95 (90, 97) 73 (12, 96) 86 (43, 98) 79 (39, 96) 0.016 PR (% positive) 9 (0, 85) 57 (0, 91) 33 (1, 95) 15 (0, 81) 0.55 p53 (% positive) 5 (0, 100) 2 (0, 40) 10 (0, 100) 1 (0, 90) 0.66 Her-2/neu (0–4)† 0 (0, 3) 2 (0, 3) 2 (0, 3) 1 (0, 3) 0.032 EGFR-tumor (0–4)† 0 (0, 0) 0 (0, 2.6) 0 (0, 0.5) 0 (0, 0) 0.18 EGFR-benign (0–4)† 2.5 (1.9, 3.5) 2.9 (0, 3.5) 2.1 (1.0, 3.5) 2.5 (0, 3.8) 0.69 Bcl-2 (0–4)† 2.0 (1.0, 3.0) 3.0 (1.0, 3.0) 3.0 (0, 3.0) 3.0 (0, 3.0) 0.29 Morphometric z-score Ϫ0.2 (Ϫ2.4, 5.9) Ϫ1.0 (Ϫ3.1, 9.2) 2.9 (Ϫ3.0, 12.0) Ϫ0.5 (Ϫ2.8, 12.4) 0.34 Angiogenesis index Ϫ0.5 (Ϫ5, 2) 2.5 (0, 3) 2.0 (Ϫ6, 3) 1.5 (Ϫ6, 3) 0.38 Abbreviations: IGFBP, IGF binding protein; SHBG, sex hormone binding globulin; TGB, thyroxine-binding globulin; ER, estrogen receptor; PR, progesterone receptor; pERK, phosphorylated extracellular regulated kinase; EGFR, epidermal growth factor receptor. * Kruskal-Wallis test. † Weighted intensity index score.

headache, and breast pain, the latter most likely related to biopsy. Three subjects who had received 10 mg of arzoxifene required hospitalization for complications from surgery, but this was not considered to be related to arzoxifene. Change in Serum and Tissue Markers for the Phase IA Trial. Baseline serum and tissue biomarker values are given in Table 5 for the 45 tissue biomarker evaluable subjects (8 control and 37 arzoxifene). For several variables (PCNA, estrogen

Fig. 1 Number of subjects on Phase IA trial who exhibit a change (prestudy to poststudy) in serum hormone or growth factor levels. The number above the bars indicate the number of subjects in whom the change was an “improvement” (in the direction that would reduce breast cancer risk); the numbers below the bars indicate subjects in whom no change was observed or the change was in an unfavorable direction. Ⅺ, controls; f, arzoxifene. IGFBP, IGF, binding protein; SHBG, sex hormone binding globulin; TBG, thyroxine-binding globulin.

excision specimens differed in that one contained only invasive Fig. 2 Number of subjects on Phase IA trial who exhibit a change cancer and the other only ductal carcinoma in situ. Median on (biopsy to re-excision) in tissue biomarkers. The numbers above the study interval was 15 days (range, 10–42 days). bars indicate the number of subjects in whom the change was an Adverse Events for the Phase IA Trial. Reported ad- “improvement” (in the direction that would reduce breast cancer risk or verse events for all 40 of the subjects during the time they were would be consistent with the presumed mechanism of action of arzoxifene); the numbers below the bars indicate subjects in whom no change receiving arzoxifene were mild (grade 2 or less) and did not was observed or the change was in an unfavorable direction. Ⅺ, con- differ between the three dose levels (Table 4). The most frequent trols; f, arzoxifene. ER, estrogen receptor; PR, progesterone receptor; adverse effects were hot flashes, mild gastrointestinal effects, AI, angiogenesis index; EGFR, epidermal growth factor receptor.

receptor, and Her-2/neu) there were differences between groups at baseline, particularly between the control and the three arzoxifene groups. It was observed that several variables exhibited changes in a direction that would be “favorable” from a standpoint of a risk factor for breast cancer (e.g., a decrease in serum IGF-I) or as consistent with the presumed mechanism of action of arzoxifene (e.g., a decrease in proliferation index). Changes in a favorable direction for a majority of subjects receiving arzoxifene were observed for serum levels of IGF-I, IGFBP-3, IGF-1:IGFBP-3 molar ratio, sex hormone binding globulin, and thyroxine-binding globulin (Fig. 1); and for tissue expression of Ki-67, Bcl-2, Her-2/neu, epidermal growth factor receptor in benign tissue, and PCNA (Fig. 2). When separated by dose group, several markers (IGF-1:IGFBP-3 molar ratio, sex hormone binding globulin, thyroxine-binding globulin, and PCNA) exhibited favorable changes in Ͼ80% of subjects at one or more dose levels Fig. 3 Changes in PCNA expression in control subjects and subjects (Table 6). For the serum markers, it appeared that the 10-mg receiving 20 mg of arzoxifene on the Phase IA trial. Two subjects on arzoxifene had high baseline values, which are plotted on a different dose was not as effective as either the 20-mg or 50-mg doses in scale. producing a favorable change. Importantly, for the tissue biomarker PCNA, all 13 of the subjects receiving 20 mg of arzoxifene exhibited a decrease in expression (Fig. 3). re-excision. In a secondary analysis, we excluded women con- Absolute changes in serum and tissue biomarker expres- sidered to have been hormonally confounded on the basis of sion are listed in Table 7 for control subjects and subjects in the estradiol levels in the premenopausal range at biopsy and in the individual drug dose groups. Compared with the no-treatment postmenopausal range at re-excision or premenopausal women control group, significant changes (pre-post study) were seen in documented to have been at a different phase of the menstrual sex hormone binding globulin, thyroxine-binding globulin, and cycle at biopsy than at re-excision. Applying these criteria, 33 PCNA for both the 20- and 50-mg dose groups using the women (4 controls and 29 arzoxifene) considered nonhormon- Wilcoxon rank-sum test (P ϭ 0.0001). ally confounded were retained for the secondary analysis. A The possibility was considered that discontinuation of hor- significant difference between nontreated controls and subjects mone replacement therapy in the peribiopsy period may have that received arzoxifene was still observed for sex hormone been responsible for some of the observed change in Ki-67 and binding globulin, thyroxine-binding globulin, and absolute PCNA expression between biopsy and re-excision for post- changes in PCNA using parametric and nonparametric analysis menopausal women. Change in Ki-67 and PCNA may also have (Table 8). Notably, the 20-mg dose provided the greatest nu- been affected in premenopausal women if the initial biopsy was merical increase in sex hormone binding globulin and the great- performed in a different portion of their menstrual cycle than est numerical decrease in PCNA.

Table 6 Proportion of evaluable subjects on the Phase IA trial demonstrating favorable changes in serum and tissue biomarker values Arzoxifene Controls 10 mg 20 mg 50 mg Total Marker (direction of favorable change) (N ϭ 8) (N ϭ 11) (N ϭ 13) (N ϭ 13) (N ϭ 37) Serum IGF-I (decrease) 50% 82% 82% 92% 85% IGFBP-3 (increase) 33% 64% 55% 50% 56% IGF-I:IGFBP-3 molar ratio (decrease) 67% 73% 100% 92% 88% SHBG (increase) 33% 46% 91% 92% 77% TBG (increase) 17% 64% 82% 92% 79% Tissue Ki-67 (decrease) 38% 55% 54% 54% 54% PCNA (decrease) 40% 70% 100% 62% 78% ER (decrease) 62% 36% 54% 39% 43% PR (increase) 38% 46% 39% 62% 49% p53 (decrease) 38% 27% 8% 15% 16% Her-2/neu (decrease) 0% 50% 70% 44% 55% EGFR-benign (decrease) 33% 67% 46% 50% 53% Bcl-2 (decrease) 38% 36% 42% 50% 43% Angiogenesis index (increase) 50% 40% 55% 50% 48% Abbreviations: IGFBP, IGF binding protein; SHBG, sex hormone binding globulin; TGB, thyroxine-binding globulin; ER, estrogen receptor; PR, progesterone receptor; EGFR, epidermal growth factor receptor.

Table 7 Changes in serum and tissue biomarker values (median and range) for 45 evaluable subjects on the Phase IA trial Arzoxifene Control 10 mg 20 mg 50 mg (N ϭ 8) (N ϭ 11) (N ϭ 13) (N ϭ 13) P value* Serum Estradiol (pg/ml) Ϫ22 (Ϫ633, 4) Ϫ6(Ϫ342, 4) Ϫ3(Ϫ186, 456) 0 (Ϫ132, 14) 0.60 Estrone (pg/ml) Ϫ9(Ϫ807, 2) Ϫ6(Ϫ165, 6) 3 (Ϫ162, 214) 0 (Ϫ53, 10) 0.17 FSH (mIU/ml) 2.4 (Ϫ3.4, 20) Ϫ2.6 (Ϫ20, 40) Ϫ6.9 (Ϫ32, 25) Ϫ1.3 (Ϫ14, 11) 0.30 LH (ng/ml) 0.8 (Ϫ33, 6) Ϫ1.7 (Ϫ28, 68) Ϫ8.1 (Ϫ26, 29) Ϫ7.5 (Ϫ43, 12) 0.50 IGF-I (ng/ml) Ϫ4.5 (Ϫ135, 37) Ϫ32 (Ϫ94, 37) Ϫ19 (Ϫ119, 13) Ϫ47 (Ϫ140, 37) 0.56 IGFBP-3 (␮g/ml) Ϫ0.25 (Ϫ0.9, 1.2) 0.27 (Ϫ0.4, 0.8) 0.12 (Ϫ0.5, 0.8) Ϫ0.02 (Ϫ0.9, 0.7) 0.56 IGF-I:IGFBP-3 (Molar ratio), ϫ 103 Ϫ5.5 (Ϫ18, 12) Ϫ12 (Ϫ22, 13) Ϫ8(Ϫ22, 0) Ϫ11 (Ϫ29, 4) 0.26 SHBG (nmol/l) Ϫ14 (Ϫ39, 4) 0 (Ϫ19, 60) 7 (Ϫ61, 24) 9 (Ϫ26, 22) 0.003 TBG (␮g/ml) Ϫ5(Ϫ9, 2) 3 (Ϫ11, 15) 3 (Ϫ11, 18) 7 (Ϫ2, 21) 0.002 Progesterone (ng/ml) 0.0 (Ϫ20, 1.4) 0.1 (Ϫ0.2, 0.3) 0.1 (Ϫ0.2, 1.3) 0.1 (Ϫ0.2, 0.3) 0.45 Tissue Ki-67 (% positive) 0 (Ϫ17, 2) 0 (Ϫ14, 14) Ϫ1(Ϫ18, 12) Ϫ1(Ϫ4, 4) 0.54 PCNA (% positive) 2 (Ϫ5,14) Ϫ3(Ϫ18, 5) Ϫ4(Ϫ27, Ϫ1) Ϫ1(Ϫ11, 8) 0.004 ER (% positive) Ϫ2(Ϫ79, 6) 1 (Ϫ52, 44) Ϫ4(Ϫ20, 38) 1 (Ϫ35, 35) 0.58 PR (% positive) Ϫ2(Ϫ50,51) 0 (Ϫ16, 48) Ϫ1(Ϫ25, 38) 7 (Ϫ10, 74) 0.27 p53 (% positive) 0 (Ϫ0.5, 0.1) 0 (Ϫ0.3, 0.1) 0 (Ϫ0.1, 0.1) 0 (Ϫ0.1, 0.1) 0.56 Her-2/neu (0–4)† 0(Ϫ0.1, 0.1) 0 (Ϫ2, 1) Ϫ0.5 (Ϫ2, 0) 0 (Ϫ2, 2) 0.12 EGFR-tumor (0–4)† 0(Ϫ0.1, 0.1) 0 (Ϫ1.8, 0) 0 (Ϫ0.5, 2.3) 0 (Ϫ0.1, 0.1) 0.57 EGFR-benign (0–4)† 0(Ϫ1.5, 0.3) Ϫ0.3 (Ϫ2.9, 1.5) 0 (Ϫ3.3, 0.9) 0 (Ϫ3.5, 0.8) 0.92) Bcl-2 (0–4)† 0(Ϫ2, 1) 0 (Ϫ2, 1) 0 (Ϫ2, 1) 0 (Ϫ2, 1) 0.41 Morphometric z-score Ϫ0.5 (Ϫ5.1, 3.0) 2.3 (Ϫ2.9, 8.0) 0.5 (Ϫ6.0, 5.0) 0.4 (Ϫ8.5, 6.6) 0.26 Angiogenesis index 0.5 (Ϫ1, 5) 0 (Ϫ4, 3) 2.0 (Ϫ2, 8) 0.5 (Ϫ3, 6) 0.60 Abbreviations: IGFBP, IGF binding protein; SHBG, sex hormone binding globulin; TGB, thyroxine-binding globulin; ER, estrogen receptor; PR, progesterone receptor; EGFR, epidermal growth factor receptor. *Kruskal-Wallis test. †Weighted intensity index score.

Upon completion of analysis of the Phase IA trial, four tissue biomarkers (PCNA and estrogen receptor). These differ- general observations were made. First, arzoxifene produced ences could impact the assessment of modulation of biomarkers little in the way of adverse events, with no evidence for a over the study interval and the subsequent analyses. Thus, any relationship to drug dose. Second, levels of serum markers conclusions regarding differences between control and drug considered as drug effect biomarkers or risk biomarkers ap- groups were viewed with caution and used only for purposes of peared to be favorably altered in a higher proportion of subjects designing the subsequent Phase IB trial. that received 20 mg or 50 mg arzoxifene than those receiving On the basis of these Phase IA results, the 20-mg dose of only 10 mg daily. Third, proliferation (PCNA expression) de- arzoxifene was selected for the subsequent randomized, placebo- creased in all of the subjects that received 20 mg of arzoxifene controlled Phase IB trial. daily. Lastly, the use of no-treatment controls (rather than subjects randomized to placebo) resulted in a group of subjects that Results of the Phase IB Trial differed from subjects receiving drug. In particular, there were Accrual and Participating Institutions. From June 5, significant differences in baseline values for two important 2000 to July 19, 2002, 81 subjects were entered onto the Phase

Table 8 Changes in serum and tissue biomarker values (median and range) for 33 subjects considered as nonhormonally confounded on the Phase IA trial Arzoxifene Control 10 mg 20 mg 50 mg (N ϭ 4) (N ϭ 7) (N ϭ 10) (N ϭ 12) P value* Median and range of change in value (Kruskal-Wallis test) SHBG (nmol/l) Ϫ7.9 (Ϫ23, 4.2) 3.9 (Ϫ17, 16) 9.9 (Ϫ61, 24) 9.0 (4, 22) 0.014 TBG (␮g/ml) Ϫ4.2 (Ϫ6.2, 1.8) 3.7 (Ϫ2.9, 15) 4.2 (Ϫ11, 18) 7.2 (3.9, 21) 0.015 PCNA (% positive) 4.7 (Ϫ4.9, 14) Ϫ3.0 (Ϫ18, 5) Ϫ4.0 (Ϫ25, Ϫ1) Ϫ0.8 (Ϫ11, 8) 0.033 Subjects demonstrating change in the “favorable” direction (Fisher’s exact test) SHBG (increase) 50% 71% 89% 100% 0.091 TBG (increase) 25% 71% 78% 100% 0.022 PCNA (decrease) 33% 67% 100% 58% 0.068 Abbreviations: SHBG, sex hormone binding globulin; TGB, thyroxine-binding globulin.

Table 9 Demographic characteristics of 74 postmenopausal subjects randomized to placebo or arzoxifene 20 mg/day on the Phase IB trial Placebo (N ϭ 27) Arzoxifene (N ϭ 47) P value Age (yrs), median (range) 61 (46, 78) 64 (44, 86) 0.15* Number non-Caucasian 2 (7%) 5 (11%) 0.27† Tumor size (cm from CRFs), median (range) 1.5 (0.4, 3.0) 1.4 (0.5, 5.0) 0.61* Number non-high-grade tumors 4 (15%) 11 (23%) 0.66† Number DCIS only (per review at KUMC) 3 (11%) 12 (26%) 0.23† Number on HRT within 30 days of biopsy 11 (41%) 23 (49%) 0.63† Time (days) between biopsy and study entry, median (range) 10 (0, 28) 11 (0, 45) 1.00* Time (days) on study agent, median (range) 20 (15, 36) 19 (9, 55) 0.21* Time (days) between biopsy and re-excision, median (range) 33 (16, 53) 29 (13, 84) 0.35* Time (days) between stopping HRT and biopsy,‡ median (range) 1 (-10, 17) 5 (-21, 28) 0.67* Abbreviations: DCIS, ductal carcinoma in situ; HRT, hormone replacement therapy; KUMC, University of Kansas Medical Center. * Wilcoxon rank-sum test. † ␹2, Fisher exact test. ‡ Negative value denotes stopped prior to biopsy; positive value denotes stopped after biopsy.

Table 10 Adverse events for 74 evaluable subjects on the analysis due to randomization to tamoxifen (5), withdrawal from Phase IB trial study before receipt of any drug (1), or receipt of both arzox- Placebo Arzoxifene ifene and tamoxifen (1). Thus, 74 subjects were evaluable for Adverse event Grade (N ϭ 27) (N ϭ 47) P value* toxicity analysis and serum pharmacodynamic analysis for the Menopausal symptoms† 1 4 (15%) 6 (13%) 0.72 placebo and arzoxifene arms. The median age of the 74 subjects 2 7 (26%) 9 (19%) evaluable for toxicity and serum pharmacodynamic parameters Headaches 1 8 (30%) 5 (11%) 0.061 was 63 years. The median baseline estradiol level was 28 pg/ml, 2 1 (4%) 1 (2%) and 46% of subjects had been taking hormone replacement Myalgias 1 6 (22%) 2 (4%) 0.0087 2 4 (15%) 2 (4%) therapy within 30 days of biopsy, including 23 of 47 (49%) Leg cramps 1 3 (11%) 0 (0) 0.045 subjects randomized to study drug and 11 of 27 (41%) subjects 2 0 (0) 1 (2%) randomized to placebo. Median time from biopsy to study entry Breast pain 1 5 (19%) 7 (15%) 0.91 was 10 days (range, 0–45), and median on study interval was 20 2 2 (7%) 3 (6%) Surgical complications 1 1 (4%) 3 (6%) 0.25 days (range, 9–55). There were no significant differences for 2 2 (8%) 0 (0%) demographics, tumor characteristics, or intervals between vari- 3 0 (0) 2 (4%) ous study events (Table 9) between the arzoxifene and placebo Overall adverse event’s 1 5 (19%) 19 (40%) 0.12 groups. worst grade Adverse Events for the Phase IB Trial. Analysis of 2 16 (59%) 17 (36%) 3 3 (11%) 3 (6%) adverse events was performed for the 74 subjects randomized to arzoxifene or placebo who received study drug. Similar to the * ␹2 Fisher exact test. † Includes hot flashes and sweating. Phase IA trial, reported adverse events were mild with 18% of subjects reporting no effects, 40% grade 1, 36% grade 2, and 6% grade 3. There were no significant differences between the arzoxifene group and the placebo group in reported adverse IB trial after signing appropriate informed consents as approved events, either globally or for specific types of toxicities (Table by the site Institutional Review Board. At least 1400 potential 10). The exception to this was the incidence of leg cramps and subjects were screened to accrue this number. Accrual distribution included four from University of Kansas Medical Center, 26 from Ohio State University, 16 from US Oncology, 15 from Table 11 Baseline serum hormones and growth factor values Desert Comprehensive Cancer Center, 11 from Loyola Univer- (median and range) for 74 evaluable subjects on the Phase IB trial sity Medical Center, 4 from Thomas Jefferson University (Phil- adelphia, PA), 2 each from Cleveland Clinic Foundation and Placebo Arzoxifene (N ϭ 27) (N ϭ 47) P value* University of Alabama, and 1 from University of California Los Angeles. The trial was initiated with a three-way randomization Estradiol (pg/ml) 35 (9, 233) 27 (8, 256) 0.23 Estrone (pg/ml) 68 (19, 1585) 63 (11, 774) 0.99 to placebo, tamoxifen 20 mg/day, or arzoxifene 20 mg/day. FSH (mIU/ml) 8 (2, 70) 9 (1, 39) 0.72 After entry of the first 14 subjects, the protocol was amended LH (ng/ml) 29 (4, 82) 29 (3, 239) 0.31 due to slow accrual. The tamoxifen arm was discontinued, and IGF-I (ng/ml) 84 (7, 305) 75 (8, 159) 0.15 ␮ a weighted randomization scheme was used such that 2 subjects IGFBP-3 ( g/ml) 3.7 (1.9, 7.7) 3.3 (1.2, 6.6) 0.16 IGF-I:IGFBP-3 molar 85 (6, 220) 73 (11, 379) 0.42 were randomized to arzoxifene for every 1 randomized to pla- ratio, ϫ 103 cebo. Of the 81 subjects accrued to the study, 5 were accrued to SHBG (nmol/l) 60 (19, 238) 93 (15, 426) 0.062 tamoxifen in the preamendment portion, and the remaining 76 Abbreviations: SHBG, sex hormone binding globulin; IGFBP, IGF were randomized between arzoxifene and placebo. Seven sub- binding protein. jects were excluded from the toxicity and pharmacodynamic * Wilcoxon rank-sum test.

Table 12 Changes in serum hormone and growth factor values (median and range) for 74 evaluable subjects on the Phase IB trial Placebo Arzoxifene (N ϭ 27) (N ϭ 47) P value* Estradiol (pg/ml) Ϫ2(Ϫ79, 46) Ϫ2(Ϫ248, 189) 0.90 Estrone (pg/ml) 2 (Ϫ183, 192) 5 (Ϫ602, 553) 0.92 FSH (mIU/ml) Ϫ0.2 (Ϫ55, 18) Ϫ0.8 (Ϫ34, 14) 0.41 LH (ng/ml) 2.5 (Ϫ34, 34) Ϫ3.3 (Ϫ29, 51) 0.031 IGF-I (ng/ml) 10 (Ϫ86, 103) Ϫ22 (Ϫ74, 70) 0.0004 IGFBP-3 (␮g/ml) 0.23 (Ϫ4.6, 1.8) Ϫ0.28 (Ϫ3.4, 3.4) 0.57 IGF-1:IGFBP-3 molar ratio, ϫ 103 8(Ϫ55, 210) Ϫ14 (Ϫ316, 221) 0.0066 SHBG (nmol/l) Ϫ4(Ϫ106, 25) 15 (Ϫ197, 96) 0.0015 Abbreviations: SHBG, sex hormone binding globulin; IGFBP, IGF binding protein. * Wilcoxon rank-sum test.

myalgias, which were more frequent in the placebo arm than the tumors were estrogen receptor ϩ and/or progesterone receptor arzoxifene arm (P ϭ 0.045 and 0.0087, respectively). ϩ at baseline, and there were malignant cells in both the pre- Surgical complications such as hematomas, seromas, and poststudy specimens. Sixteen subjects were not evaluable wound dehiscence, and infection were no more common in the for tissue-based biomarkers, because there was no residual tu- arzoxifene group than in the placebo group (12% for both). Five mor available for analysis in the biopsy and/or re-excision serious adverse events (hypotension, hemorrhage, sinus tachy- specimen (11 subjects), because the biopsy was both estrogen- cardia with arrhythmia, wound infection, and acute otitis) were receptor and progesterone-receptor negative (3 subjects) by im- reported in 5 subjects after planned re-excision surgery and were munohistochemistry performed at University of Kansas Medical not considered to be study agent-related. The sinus tachycardia Center or because Ki-67 values were not available for both and the wound infection events were in subjects randomized to biopsy and re-excision specimens (2 subjects). Of the 58 sub- arzoxifene; the other three events were in subjects randomized jects evaluable for modulation of Ki-67, 37 were randomized to to placebo. arzoxifene and 21 to placebo; for 54 subjects with pre-post Change in Serum Pharmacodynamic Parameters for values for PCNA, 35 were randomized to arzoxifene and 19 to the Phase IB Trial. Baseline values for serum pharmacody- placebo. Distributions of demographic characteristics (Table 13) namic parameters for the 74 placebo and arzoxifene subjects are were similar to the larger (74 subjects) group evaluated for summarized in Table 11. There was no significant difference in toxicity. Median age was 64 (range, 44–86) years; 42% had baseline values between the two groups. Changes in estradiol, been on hormone replacement therapy within 30 days of biopsy; estrone, FSH, LH, IGF-I, IGFBP-3, IGF-1:IGFBP-3 molar ratio, median time from biopsy to study entry was 10 (range, 0–45) and sex hormone binding globulin for placebo and arzoxifene days; and median on study interval was 20 (range, 9–55) days. groups are given in Table 12. Compared with placebo, arzox- Forty seven (81%) had invasive cancers evaluated, and 11 ifene-treated subjects had significantly greater increases in sex (19%) ductal carcinoma in situ; 45 (78%) had non-high-grade hormone binding globulin (P ϭ 0.0015); and greater decreases tumors. in LH (P ϭ 0.03), IGF-I (P ϭ 0.0004), and IGF-I:IGFBP-3 Distributions of baseline serum hormone levels (Table 14) molar ratio (P ϭ 0.0066). were also similar to the larger (74 subjects) group evaluated for Change in Tissue Biomarkers for the Phase IB Trial. serum hormone pharmacodynamics. There were no significant Subjects randomized to arzoxifene or placebo were evaluable differences between the arzoxifene and placebo groups for base- for the primary study endpoint of change in proliferation if the line biomarker expression of any serum or tissue biomarker,

Table 13 Demographic characteristics of 58 postmenopausal subjects randomized to placebo or arzoxifene 20 mg/day and evaluable for primary objective on the Phase IB trial Placebo (N ϭ 21) Arzoxifene (N ϭ 37) P value Age (years) Median (range) 62 (46–78) 64 (44, 86) 0.24* Number non-Caucasian 2 (9%) 3 (8%) 0.54† Tumor size (cm from CRFs) Median (range) 1.5 (0.7, 3.0) 1.5 (0.5, 5.0) 0.89* Number non-high-grade tumors 18 (86%) 27 (73%) 0.37† Number DCIS only (per review at KUMC) 3 (14%) 8 (22%) 0.73† Number on HRT within 30 days of biopsy 7 (33%) 18 (49%) 0.29† Time (days) between biopsy and study entry Median (range) 10 (0, 28) 9 (0, 45) 0.58* Time (days) on study agent Median (range) 21 (15, 36) 20 (9, 55) 0.26* Time (days) between biopsy and re-excision Median (range) 31 (16, 53) 29 (13, 84) 0.64* Time (days) between stopping HRT and biopsy‡ Median (range) 1 (0, 15) 6.5 (Ϫ21, 28) 0.48* Abbreviations: DCIS, ductal carcinoma in situ; HRT, hormone replacement therapy; KUMC, University of Kansas Medical Center. * Wilcoxon rank-sum test. † ␹2, Fisher exact test. ‡ Negative value denotes stopped before biopsy; positive value denotes stopped after biopsy.

Table 14 Values (median, range) for baseline serum and tissue evaluable subjects had been on hormone replacement therapy parameters of 58 subjects evaluable for the primary objective on the at the time of biopsy or within 30 days before biopsy but Phase IB trial stopped before study entry (median of 4 days after biopsy). Placebo Arzoxifene These included 7 of 21 (33%) subjects in the placebo group ϭ ϭ (N 21) (N 37) P value* and 18 of 37 (49%) subjects in the arzoxifene group. As Serum anticipated, differences were observed in Ki-67 modulation Estradiol (pg/ml) 34 (9, 232) 28 (6, 200) 0.25 between subjects on hormone replacement therapy versus Estrone (pg/ml) 60 (24, 1585) 67 (11, 774) 0.63 FSH (mIU/ml) 8 (2, 70) 8 (2, 39) 0.42 those not on hormone replacement therapy within 30 days of LH (ng/ml) 28 (4–82) 29 (6–120) 0.23 biopsy. Subjects not on hormone replacement therapy exhib- IGF-I (ng/ml) 78 (7–305) 75 (8–159) 0.34 ited a median relative decrease in the proportion of cells ␮ IGFBP-3 ( g/ml) 3.6 (1.9, 7.7) 3.6 (1.6, 5.5) 0.30 expressing Ki-67 of 8% for arzoxifene versus an increase of IGF-I:IGFBP-3 molar 82 (6, 220) 71 (11, 379) 0.50 ratio, ϫ 103 3% for placebo and a median relative decrease in the pro- SHBG (nmol/l) 60 (19, 238) 93 (14, 426) 0.12 portion of cells expressing PCNA of 25% for arzoxifene Tissue versus an increase of 3% for placebo. Subjects who had Ki-67 (% positive) 6 (2, 41) 7 (0, 39) 0.97 received hormone replacement therapy but then discontinued PCNA (% positive) 5 (1, 41) 10 (0, 44) 0.27 use exhibited a relative decrease in the median proportion of ER (% positive) 95 (6, 99) 95 (15, 99) 0.90 PR (% positive) 63 (0, 100) 66 (0, 100) 0.72 cells expressing Ki-67 of 53% for arzoxifene versus 76% for p53 (% positive) 0 (0, 84) 0 (0, 38) 0.51 placebo and a relative decrease in the median proportion of pERK (% positive) 7 (0, 95) 8 (0, 95) 0.52 cells expressing PCNA of 50% for arzoxifene versus 56% for Her-2/neu (0–4)† 0.9 (0, 2.9) 0.8 (0, 3.6) 0.26 placebo. These results would appear to indicate a strong EGFR-tumor (0–4)† 0 (0, 3.0) 0 (0, 1.5) 0.26 EGFR-benign (0–4)† 2.5 (0, 3.5) 2.0 (0, 3.0) 0.11 antiproliferative effect of stopping hormone replacement Bcl-2 (0–4)† 3 (.5, 4) 2 (0, 4) 0.07 therapy independent of any effect of arzoxifene. IGFR (0–4)† 3 (1, 4) 3 (1.5, 4) 0.33 TNF-␣ (0–4)† 3 (2, 4) 3 (1, 4) 0.039‡ MVD (no. vessels/ 18 (6, 8) 18 (0, 32) 0.69 DISCUSSION 5 HPF) Reduction in the proportion of tumor cells expressing Abbreviations: IGFBP, IGF binding protein; SHBG, sex hormone Ki-67 after 2–3 weeks of tamoxifen has been correlated with binding globulin; ER, estrogen receptor; PR, progesterone receptor; later clinical response in women with breast cancer (12, 15). pERK, phosphorylated extracellular regulated kinase; EGFR, epidermal growth factor receptor; IGFR, IGF receptor; TNF, tumor necrosis factor; Several European studies have used preoperative models similar MVD, microvessel density; HPF, high powered fields. to our Phase IA/IB trial and found reductions in Ki-67 after 2–4 * Wilcoxon rank-sum test. weeks of treatment with a selective estrogen receptor modulator † Weighted intensity index score. (14, 16). ‡ Change in distribution only, median values are identical. In the trial reported by Decensi et al., a 15% (95% confidence interval, 0–24%) median relative reduction from baseline in the proportion of cells expressing Ki-67 was observed for 120 including the proliferation indices. However, it should be noted estrogen receptor ϩ subjects randomized to one of three doses that values for proliferation indices were low. The median of tamoxifen for 4 weeks, whereas a relative increase of 13% values for proportion of cells expressing Ki-67 and PCNA at (95% confidence interval, Ϫ3.3–33%) was observed in 29 es- baseline were of 6% and 5% for placebo, and 7% and 10% for trogen receptor ϩ nonrandomized controls (16). Dowsett et al. arzoxifene, respectively. (14) reported a 21% median relative reduction from baseline in Table 15 compares the changes in serum and tissue biomar- the proportion of cells expressing Ki-67 for 42 subjects random- kers for the 58 evaluable subjects. Significant differences were ized to 2 weeks of 60 mg/day raloxifene versus a median observed between the placebo and arzoxifene groups for serum relative increase of 7% in 36 subjects randomized to placebo. LH, sex hormone binding globulin, IGF-I, the molar ratio of Dowsett et al. (13) also reported a 35% mean relative reduction IGF-I:IGFBP-3, and tumor estrogen receptor (P Յ 0.01). No from baseline in the proportion of cells expressing Ki-67 for 30 significant differences were observed between placebo and ar- subjects randomized to 2–3 weeks of idoxifene versus a6% zoxifene groups for the main study endpoint, tumor cell prolif- mean relative increase in 27 subjects randomized to placebo. eration, despite decreases in both Ki-67 and PCNA in the The trial reported by Decensi et al. (16) differed from ours in arzoxifene group. The median change in the proportion of cells that three times as many subjects were entered into the study and expressing Ki-67 was Ϫ1% for placebo and Ϫ2% for arzox- were treated for twice as long. The two studies reported by ifene; the corresponding relative changes were Ϫ20% and Dowsett et al. (13, 14) differed from ours in that eligible women Ϫ25%, respectively. The median change in the proportion of could not have received hormone replacement therapy for at cells expressing PCNA was 0% for placebo and Ϫ3% for least 3 months before initial biopsy. arzoxifene; the corresponding relative changes were Ϫ23% and In our IA arzoxifene study, a decrease in Ki-67 from Ϫ40%, respectively. baseline was observed in only 54% of subjects, whereas PCNA Similar to the Phase IA trial, a secondary analysis was was decreased in Ͼ80% of subjects at the higher dose levels. performed to assess the potential confounding effects on PCNA is more likely than Ki-67 to be expressed in cells in early

proliferation of stopping the hormone replacement therapy in G1 and/or in cells that have traversed the cell cycle recently but the peribiopsy period. Twenty-five of the 58 tissue biomarker are currently in G0 (38–40). Thus, when proliferation is low, as

Table 15 Change in serum and tissue biomarkers of 58 women evaluable for the primary objective on the Phase IB trial Placebo (N ϭ 21) Arzoxifene (N ϭ 37) Biomarker median (range) median (range) P value* Serum Estradiol (pg/ml -2.2 (-79, 46) -1.6 (-141, 189) 0.52 Estrone (pg/ml) 2.2 (-183,192) 4.9 (-602, 553) 0.63 FSH (mIU/ml) 0.1 (-55,18) -1.2 (-34, 14) 0.17 LH (ng/ml) 2.5 (-34, 34) -3.7 (-29, 11) 0.0166 IGF-I (ng/ml) 10 (-86, 103) -25 (-74, 62) 0.0015 IGFBP-3 (␮g/ml) 0.23 (-4.6, 1.7) -0.3 (-3.4, 3.4) 0.96 IGF-I:IGFBP-3 molar ratio, ϫ103 9 (-55, 210) -14 (-315, 77) 0.0049 SHBG (nmol/l) -2.2 (-106, 25) 17.2 (-197, 96) 0.0049 Tissue Ki-67(% positive) -1.1 (-36, 13) -1.5 (-24, 7) 0.55 PCNA(% positive) 0 (-36, 17) -3 (-20, 5) 0.12 ER (% positive) 0 (-45, 30) -6 (-99, 45) 0.0068 PR (% positive) -1 (-92, 57) -1 (-91, 65) 0.92 p53 (% positive) 0 (-83, 11) 0 (-27, 3) 0.57 pERK (% positive) 0 (-59, 30) -5 (-75, 70) 0.14 Her-2/neu (0–4)† -0.20 (-170, 60) -0.10 (-130, 110) 0.44 EGFR tumor (0–4)† 0 (-0.75, 0) 0 (0, 0) 0.19 EGFR benign (0–4)† 0 (-2.0, 1.5) 0 (-2.5, 1.8) 0.78 Bcl-2 (0–4)† 0 (-2.0, 1.5) 0 (-2.0, 2.0) 0.89 IGFR (0–4)† 0 (-1.0, 1.0) 0 (-2.5, 1.5) 0.19 TNF-␣ (0–4)† 0 (-1.0, 0) 0 (-20, 1.0) 0.73 MVD (no. vessels/5 HPF) -3.7 (-11, 12) -1.1 (-23, 13) 0.15 Abbreviations: IGFBP, IGF binding protein; SHBG, sex hormone binding globulin; ER, estrogen receptor; PR, progesterone receptor; pERK, phosphorylated extracellular regulated kinase; EGFR, epidermal growth factor receptor; IGFR, IGF receptor; TNF, tumor necrosis factor; MVD, microvessel density; HPF, high powered fields. * Wilcoxon rank-sum test. † Weighted intensity index score.

observed in the predominately well-differentiated tumors in this tween core needle biopsy and definitive surgery (re-excision, clinical model, PCNA may be a more sensitive indicator of lumpectomy, or mastectomy) for invasive breast cancer. They proliferation than Ki-67 (41, 42). Unfortunately, PCNA detec- found that there was a decrease in Ki-67 expression associated tion varies with length of fixation and processing conditions. with discontinuation of hormone replacement therapy that was Moreover, assessment of positive staining is more susceptible to statistically significant when compared with patients that had interpretive variance than Ki-67 (43). Although the arzoxifene either not been using hormone replacement therapy before bi- dose for our Phase IB trial was selected on the basis of PCNA, opsy or who had continued to use hormone replacement therapy Ki-67 is considered to be more reproducible and consistent in its until definitive surgery. This effect was observed for tumors that pattern of staining (42, 44, 45) and, thus, both were retained as were estrogen receptor ϩ but not for tumors that were estrogen measures of proliferation. In the Phase IB trial, the baseline receptor Ϫ. expression of Ki-67 and PCNA were similar, both in the placebo For those women in our study not on hormone replacement group and in the arzoxifene group. therapy previously, the 3% (Ki-67) or 17% (PCNA) median For women randomized to arzoxifene in the Phase IB relative increase for placebo and the 8% (Ki-67) or 30% trial, we observed a median relative reduction of 25% in the (PCNA) median relative decrease for arzoxifene are similar to proportion of tumor cells expressing Ki-67, similar to that proliferation effects observed in the United Kingdom trials of reported for tamoxifen, raloxifene, and idoxifene, as well as raloxifene and idoxifene. On the other hand, subjects who had a 46% reduction in PCNA. However, unlike the trials re- received hormone replacement therapy within 30 days of biopsy ported by Dowsett et al. (13, 14), a median relative reduction exhibited a dramatic reduction in the proportion of cells express- of 20% in Ki-67 and 23% in PCNA was also observed in the ing Ki-67 or PCNA, regardless of whether randomized to pla- randomized placebo group. Thus, for our primary endpoint cebo or to arzoxifene. This implies that future studies using the proliferation, there was no significant change between 20 preoperative model should not enroll women with estrogen mg/day arzoxifene and placebo. Subgroup analysis indicates receptor ϩ tumors who have been taking hormone replacement that the lack of significant reduction in proliferation index in therapy before biopsy if proliferation is to be used as an end- women randomized to arzoxifene versus those randomized to point and if the hormone replacement therapy is to be discon- placebo may have been in part secondary to the confounding tinued between biopsy and re-excision. Although our Phase IB effects of discontinuance of hormone replacement therapy in trial was initially designed to exclude both premenopausal the peribiopsy period before re-excision. women as well as postmenopausal women who had received Prasad et al. (46) have reported recently on the effects of hormone replacement therapy recently, accrual problems discontinuing hormone replacement therapy in the interval be- prompted us to amend the study to allow entry of postmeno-

pausal women on hormone replacement therapy at the time of our randomized, double-blind Phase IB trial. Failure to demon- biopsy. strate significant reduction in proliferation relative to placebo Another factor that may contribute variability and, thus, may be secondary to the small sample size, the confounding affect the evaluation of modulation due to a drug effect is the effects of discontinuing hormone replacement therapy in a high duration of exposure to study agent. This variation is inherent in proportion of study subjects, or other factors. We were able to this clinical model where the scheduling of re-excision surgery demonstrate favorable modulation of tumor estrogen receptor could not be dictated by the study. Rather, the timing of surgery expression and serum LH, IGF-I:IGFBP-3 ratio, and sex hor- was at the discretion of the treating physician and was planned mone binding globulin relative to placebo. Favorable modula- before study entry. To participate, this planned interval could be tion of these biomarkers and a favorable toxicity profile in Phase no shorter than 2 weeks and no longer than 6 weeks. The actual II treatment trials continues to make this agent an excellent time on study agent ranged from 9 to 55 days for the 58 subjects candidate for additional study in the prevention setting, either evaluable for Ki-67. Three subjects received agent for Ͻ15 days alone or in combination with other agents. A Phase II chemo- and 5 subjects received agent for Ͼ30 days. There was no prevention trial in 200 women at high risk for development of difference for the time on study agent between the placebo breast cancer will be completed in 2004. The primary endpoint (median 21 days) and arzoxifene (median 20 days) groups (Table 13). Whereas it might seem desirable to have a standard, is modulation of cytomorphology in specimens obtained by longer interval on study agent, there was no evidence to suggest random periareolar fine needle aspirations at baseline and after that a longer exposure to arzoxifene resulted in any greater 6 months of study agent. modulation of proliferation. Serum IGF-I:IGFBP-3 ratio and non-sex hormone binding REFERENCES globulin bound levels of estradiol, as well as free testosterone have been reported as risk factors for breast cancer (19, 47). LH, 1. Sato M, Turner CH, Wang T, Adrian MD, Rowley E, Bryant HU. LY353381.HCl: a novel raloxifene analog with improved SERM po- which stimulates the conversion of cholesterol to pregnenolone tency and efficacy in vivo. J Pharmacol Exp Ther 1998;287:1–7. as the first step in steroidogenesis, has been found to be elevated 2. Suh N, Glasebrook AL, Palkowitz AD, et al. Arzoxifene, a new in breast cancers (48). High levels of sex hormone binding selective estrogen receptor modulator for chemoprevention of experi- globulin, which may result in decreased levels of bioavailable mental breast cancer. Cancer Res 2001;61:8412–5. estradiol and testosterone, are associated with reduced breast 3. Schafer JM, Lee ES, Dardes RC, et al. Analysis of cross-resistance of cancer risk (20). The reductions in serum LH, IGF-I, and IGF- the selective estrogen receptor modulators arzoxifene (LY353381) and I:IGFBP-3 molar ratio and the increase in sex hormone binding LY117018 in tamoxifen-stimulated breast cancer xenografts. Clin Can- globulin observed with arzoxifene are similar to those reported cer Res 2001;7:2505–25. for tamoxifen (21, 49) and are probably the result of partial 4. Lavinsky RM, Jepsen K, Heinzel T, et al. Diverse signaling pathways modulate nuclear receptor recruitment of N-CoR and SMRT complexes. estrogen agonist properties of arzoxifene on the central nervous Proc Natl Acad Sci USA 1998;95:2920–5. system and liver metabolism (50, 51). Change in serum hor- 5. Graham JD, Bain DL, Richer JK, Jackson TA, Tung L, Horwitz KB. mones in our trial of women with ductal carcinoma in situ or Thoughts on tamoxifen resistant breast cancer. Are coregulators the small invasive cancers are similar to those reported in Phase I answer or just a red herring? J Steroid Biochem Mol Biol 2000;74: and II treatment trials of arzoxifene in women with metastatic 255–9. disease (9, 11). 6. Ke HZ, Chen HK, Simmons HA, et al. Comparative effects of Both intraepithelial neoplasia and breast cancer are often droloxifene, tamoxifen, and estrogen on bone, serum cholesterol, and associated with an increase in the proportion of cells expressing uterine histology in the ovariectomized rat model. Bone 1997;20: 31–9. estrogen receptor (52–54). The significant (P ϭ 0.0068) reduc- 7. Suh N, Lamph WW, Glasebrook AL, et al. Prevention and treatment tion in estrogen receptor expression that we observed for arzox- of experimental breast cancer with the combination of a new selective ifene has also been reported for idoxifene (13) but not consis- estrogen receptor modulator, arzoxifene, and a new rexinoid, LG tently for tamoxifen (12, 55, 56). We also observed that for 100268. Clin Cancer Res 2002;8:3270–5. those subjects not on hormone replacement therapy in the 8. Ferguson L, Ni L, Knadler MP. Pharmacokinetics of LY353381 in peribiopsy period, there were significant (P ϭ 0.009) differ- healthy postmenopausal women after multiple dose oral adminstration. ences in both absolute change and relative change for placebo Proc Am Assoc Pharmaceut Sci 1999;A2015 (abstract). (medians of 0%) versus arzoxifene (median 5% decreases). For 9. Munster PN, Buzdar A, Dhingra K et al. Phase I study of a third- generation selective estrogen receptor modulator, LY353381.HCL, in subjects on hormone replacement therapy in the peribiopsy metastatic breast cancer. J Clin Oncol 2001;19:2002–9. period, the differences in change in estrogen receptor expression 10. Baselga J, Llombart-Cussat A, Bellet M, et al. Double-blind ran- between placebo and arzoxifene were not statistically signifi- domized phase II study of a selective estrogen receptor modulator in cant, again underscoring the powerful effect that stopping hor- patients with locally advanced or metastatic breast cancer. Breast Can- mone replacement therapy has upon breast tissue biomarker cer Res Treat 1999;57:A25. expression. Importantly, unlike what has been reported for ta- 11. Buzdar A, O’Shaughnessy JA, Booser DJ, et al. Phase II, random- moxifen (57), no increase was observed for progesterone recep- ized, double-blind study of two dose levels of arzoxifene in patients with tor expression after arzoxifene administration, which might in- locally advanced or metastatic breast cancer. J Clin Oncol 2003;21: 1007–14. dicate less likelihood of a partial estrogen agonist affect on 12. Makris A, Powles TJ, Allred DC, et al. Changes in hormone breast tissue. receptors and proliferation markers in tamoxifen treated breast cancer In summary, we did not demonstrate a significant reduction patients and the relationship with response. Breast Cancer Res Treat in tumor cell proliferation with arzoxifene relative to placebo in 1998;48:11–20.

13. Dowsett M, Dixon JM, Horgan K, Salter J, Hills M, Harvey E. antagonist-occupied steroid receptors. J Steroid Biochem Mol Biol Antiproliferative effects of idoxifene in a placebo-controlled trial in 1999;69:45–50. primary human breast cancer. Clin Cancer Res 2000;6:2260–7. 34. Kahlert S, Nuedling S, van Eickels M, Vetter H, Meyer R, Grohe C. 14. Dowsett M, Bundred NJ, Decensi A, et al. Effect of raloxifene on Estrogen receptor alpha rapidly activates the IGF-1 receptor pathway. breast cancer cell Ki67 and apoptosis: a double-blind, placebo-con- J Biol Chem 2000;275:18447–53. trolled, randomized clinical trial in postmenopausal patients. Cancer 35. Purohit A, Newman SP, Reed MJ. The role of cytokines in regu- Epidemiol Biomark Prev 2001;10:961–6. lating estrogen synthesis: implications for the etiology of breast cancer. 15. Chang J, Powles TJ, Allred DC, et al. Prediction of clinical outcome Breast Cancer Res 2002;4:65–6. from primary tamoxifen by expression of biologic markers in breast 36. Oesterreich S, Zhang P, Guler RL, et al. Re-expression of estrogen cancer patients. Clin Cancer Res 2000;6:616–21. receptor alpha in estrogen receptor alpha-negative MCF-7 cells restores 16. Decensi A, Robertson C, Viale G, et al. A randomized trial of both estrogen and insulin-like growth factor-mediated signaling and low-dose tamoxifen on breast cancer proliferation and blood estrogenic growth. Cancer Res 2001;61:5771–7. biomarkers. J Natl Cancer Inst 2003;95:779–90. 37. Osborne CK, Bardou V, Hopp TA, et al. Role of the estrogen 17. Fabian CJ, Kimler BF, Elledge RM, Grizzle WE, Beenken SW, receptor coactivator AIB1 (SRC-3) and HER-2/neu in tamoxifen resist- Ward JH. Models for early chemoprevention trials in breast cancer. ance in breast cancer. J Natl Cancer Inst 2003;95:353–61. Hematol/Oncol Clin N Am 1998;12:993–1017. 38. Gerdes J, Schwab U, Lemke H, Stein H. Production of a mouse 18. Singletary SE, Atkinson EN, Hoque A, et al. Phase II clinical trial monoclonal antibody reactive with a human nuclear antigen associated of N-(4-hydroxyphenyl)retinamide and tamoxifen administration before with cell proliferation. Int J Cancer 1983;31:13–20. definitive surgery for breast neoplasia. Clin Cancer Res 2002;8:2835– 39. van Dierendonck JH, Keijzer R, van de Velde CJ, Cornelisse 42. CJ. Nuclear distribution of the Ki-67 antigen during the cell cycle: 19. Hankinson SE, Willett WC, Colditz GA, et al. Circulating concen- comparison with growth fraction in human breast cancer cells. Cancer trations of insulin-like growth factor-I and risk of breast cancer. Lancet Res 1989;49:2999–3006. 1998;351:1393–6. 40. van Dierdendonck, Wijsman JH, Keijzer A, van de Velde CJH, 20. The Endogenous Hormones and Breast Cancer Collaborative Cornelisse CJ. Cell-cycle related staining patterns of anti-proliferative Group. Endogenous sex hormones and breast cancer in postmenopausal cell nuclear antigen monoclonal antibodies. Comparison with BrdUrd women: reanalysis of nine prospective studies. J Natl Cancer Inst labeling and Ki-67 staining. Am J Pathol 1991;138:1165–72. 2002;94:606–16. 41. Shrestha P, Yamada K, Wada T, et al. Proliferating cell nuclear 21. Lonning PE, Johannessen DC, Lien EA, Ekse D, Fotsis T, Adler- antigen in breast lesions: correlation of c-erbB-2 oncoprotein and EGF creutz H. Influence of tamoxifen on sex hormones, gonadotrophins and receptor and its clinicopathological significance in breast cancer. Vir- sex hormone binding globulin in postmenopausal breast cancer patients. chows Arch A Pathol Anat Histopathol 1992;421:193–202. J Steroid Biochem Mol Biol 1995;52:491–6. 42. Tuccari G, Rizzo A, Muscara M, Giuffre G, Barresi G. PCNA/ 22. Ciocca DR, Elledge R. Molecular markers for predicting response cyclin expressin in breast carcinomas: its relationships with Ki-67, ER, to tamoxifen in breast cancer patients. Endocrine 2000;13:1–10. PgR immunostaining and clinico-pathologic aspects. Pathologica 1993; 23. Mass R. The role of HER-2 expression in predicting response to 85:47–55. therapy in breast cancer. Semin Oncol 2000;27(Suppl 11):46–52. 43. Yu CC, Dublin EA, Camplejohn RS, Levison DA. Optimization of 24. Mehta M, Kyshtoobayeva A, Kurosaki T, et al. Independent asso- immunohistochemical staining of proliferating cells in paraffin sections ciation of angiogenesis index with outcome in prostate cancer. Clin of breast carcinoma using antibodies to proliferating cell nuclear antigen Cancer Res 2001;7:81–9. and the Ki-67 antigen. Anal Cell Pathol 1995;9:45–52. 25. Long BJ, Lu Q, Brodie A. Expression and regulation of vascular 44. Biesterfeld S, Kluppel D, Koch R, et al. Rapid and prognostically endothelial growth factor (VEGF) in human breast cancer cells. Proc valid quantification of immunohistochemical reactions by immunohis- Am Assoc Cancer Res 1997;38:572. tometry of the most positive tumour focus. A prospective follow-up 26. Kato T, Kimura T, Ishii N, et al. The methodology of quantitation study on breast cancer using antibodies against MIB-1, PCNA, ER, and of microvessel density and prognostic value of neovascularization as- PR. J Pathol 1998;185:25–31. sociated with long-term survival in Japanese patients with breast cancer. 45. Keshgegian AA, Cnaan A. Proliferation markers in breast carci- Breast Cancer Res Treat 1999;53:19–31. noma. Mitotic figure count, S-phase fraction, proliferating cell nuclear 27. Keen JC, Dixon JM, Miller EP, et al. The expression of Ki-S1 and antigen, Ki-67 and MIB-1. Am J Clin Pathol 1995;104:42–9. bcl-2 and the response to primary tamoxifen therapy in elderly patients 46. Prasad R, Boland GP, Cramer A, Anderson E, Knox WF, Bundred with breast cancer. Breast Cancer Res Treat 1997;44:123–33. NJ. Short-term biologic response to withdrawal of hormone replacement 28. Suganuma M, Okabe S, Sueoka E, et al. A new process of cancer therapy in patients with invasive breast carcinoma. Cancer 2003;98: prevention mediated through inhibition of tumor necrosis factor alpha 2539–46. expression. Cancer Res 1996;56:3711–5. 47. Cauley JA, Lucas FL, Kuller LH, Stone K, Browner W, Cum- 29. Fujiki H, Sueoka E, Komori A, Suganuma M. Tumor promotion mings S. Elevated serum estradiol and testosterone concentrations and TNF-␣ expression by the okadaic acid class tumor promoters. are associated with a high risk for breast cancer. Ann Intern Med Environ Carcino Ecotox Revs 1997;C15:1–40. 1999;130:270–7. 30. Nickerson T, Yang X-F, Pollak M, Huynh H. Post-transcriptional 48. Silva EG, Mistry D, Li D, et al. Elevated luteinizing hormone in regulation of insulin-like growth factor I receptor by the pure antiestro- serum, breast cancer tissue, and normal breast tissue from breast cancer gen ICI 182,780. Proc Am Assoc Cancer Res 1996;37:221. patients. Breast Cancer Res Treat 2002;76:125–30. 31. Grizzle WE, Meyers RB, Oelschlager DK. Prognostic biomarkers in 49. Decensi A, Gandini S, Guerrieri-Gonzaga A, et al. Effect of blood breast cancer: factors affecting immunohistochemical evaluation. The tamoxifen concentrations on surrogate biomarkers in a trial of dose Breast J 1995;1:243–50. reduction in healthy women. J Clin Oncol 1999;17:2633–8. 32. Bacus JW, Boone CW, Bacus JV, et al. Image morphometric 50. Littleton-Kearney MT, Ostrowski NL, Cox DA, Rossberg MI, Hurn nuclear grading of intraepithelial neoplastic lesions with applications to PD. Selective estrogen receptor modulators: tissue actions and potential cancer chemoprevention trials. Cancer Epidemiol Biomark Prev 1999; for CNS protection. CNS Drug Rev 2002;8:309–30. 8:1087–94. 51. Osborne CK, Zhao H, Fuqua SA. Selective estrogen receptor mod- 33. Takimoto GS, Graham JD, Jackson TA, et al. Tamoxifen resistant ulators: structure, function, and clinical use. J Clin Oncol 2000;18: breast cancer: coregulators determine the direction of transcription by 3172–86.

52. Khan SA, Rogers MA, Khurana KK, Meguid MM, Numann PJ. tamoxifen on Ki67 labelling index in human breast tumours and its Estrogen receptor expression in benign breast epithelium and breast relationship to oestrogen and progesterone receptor status. Br J Cancer cancer risk. J Natl Cancer Inst 1998;90:37–42. 1993;67:606–11. 53. Allred DC, Mohsin SK, Fuqua SA. Histological and biological 56. Fernando IN, Titley JC, Powles TJ, et al. Measurement of S-phase evolution of human premalignant breast disease. Endocr Relat Cancer fraction and ploidy in sequential fine-needle aspirates from primary 2001;8:47–61. human breast tumours treated with tamoxifen. Br J Cancer 1994;70: 54. Shaaban AM, Sloane JP, West CR, Foster CS. Breast cancer risk in 1211–6. usual ductal hyperplasia is defined by estrogen receptor-alpha and Ki-67 57. Noguchi S, Miyauchi K, Nishizawa Y, Koyama H. Induction of expression. Am J Pathol 2002;160:597–604. progesterone receptor with tamoxifen in human breast cancer with 55. Clarke RB, Laidlaw IJ, Jones LJ, Howell A Anderson E. Effect of special reference to its behavior over time. Cancer 1998;61:1345–9.

Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2004 American Association for Cancer Research. Breast Cancer Chemoprevention Phase I Evaluation of Biomarker Modulation by Arzoxifene, a Third Generation Selective Estrogen Receptor Modulator

Carol J. Fabian, Bruce F. Kimler, Julie Anderson, et al.

Clin Cancer Res 2004;10:5403-5417.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/10/16/5403

Cited articles This article cites 55 articles, 20 of which you can access for free at: http://clincancerres.aacrjournals.org/content/10/16/5403.full#ref-list-1

Citing articles This article has been cited by 13 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/10/16/5403.full#related-urls

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/10/16/5403. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.