Preclinical Activity of Abemaciclib Alone Or in Combination With

Published OnlineFirst February 26, 2018; DOI: 10.1158/1535-7163.MCT-17-0290

Small Molecule Therapeutics Molecular Cancer Therapeutics Preclinical Activity of Abemaciclib Alone or in Combination with Antimitotic and Targeted Therapies in Breast Cancer Neil O'Brien1, Dylan Conklin1, Richard Beckmann2, Tong Luo1, Kevin Chau1, Josh Thomas1, Ann Mc Nulty2, Christophe Marchal2, Ondrej Kalous1, Erika von Euw1, Sara Hurvitz1, Colleen Mockbee2, and Dennis J. Slamon1

Abstract

The cyclinD:CDK4/6:Rb axis is dysregulated in a variety of inhibition with abemaciclib combined with antimitotic agents, human cancers. Targeting this pathway has proven to be a suc- both in vitro and in vivo, did not antagonize the effect of either þ cessful therapeutic approach in ER breast cancer. In this study, agent. Finally, we identified a set of Rb/E2F-regulated genes that in vitro and in vivo preclinical breast cancer models were used to consistently track with growth inhibitory response and constitute investigate the expanded use of the CDK4/6 inhibitor, abemaci- potential pharmacodynamic biomarkers of response to abe- clib. Using a panel of 44 breast cancer cell lines, differential maciclib. Taken together, these data represent a comprehensive sensitivity to abemaciclib was observed and was seen predomi- analysis of the preclinical activity of abemaciclib, used alone or þ þ þ nately in the luminal ER /HER2 and ER /HER2 subtypes. in combination, in human breast cancer models. The subtypes However, a subset of triple-negative breast cancer (TNBC) cell most likely to respond to abemaciclib-based therapies can be lines with intact Rb signaling were also found to be responsive. identified by measurement of a specific set of biomarkers Equivalent levels of tumor growth inhibition were observed in associated with increased dependency on cyclinD:CDK4/6:Rb þ þ þ ER /HER2 ,ER /HER2 as well as biomarker selected TNBC signaling. These data support the clinical development of xenografts in response to abemaciclib. In addition, abemaciclib abemaciclib as monotherapy or as a combination partner in þ þ þ combined with hormonal blockade and/or HER2-targeted ther- selected ER /HER2 ,HER2 /ER , and TNBCs. Mol Cancer Ther; apy induced significantly improved antitumor activity. CDK4/6 17(5); 897–907. 2018 AACR.

Introduction The cyclin-dependent kinases 4 and 6 (CDK4/6) are key reg- ulators of the G –S-phase restriction checkpoint and pharmaco- Advances in our understanding of the molecular diversity of 1 logically targeting these proteins in combination with hormonal human breast cancer have led to the development of specific, blockade has been shown to provide significant therapeutic molecularly targeted therapies with subsequent improvement in þ benefit to patients with advanced ER /HER2 breast cancer clinical outcomes for patients with hormone receptor positive þ (13–15). In ER-dependent tumors, activated ER signaling leads (ER ) and human epidermal growth factor receptor-2 positive þ to increased transcription and synthesis of cyclin D1, promoting (HER2 ) disease (1–5). Despite these advances, a number of formation of activating complexes with CDK4/6 (16). Phosphor- patients with these subtypes who receive appropriate therapies ylation of the retinoblastoma protein (Rb) by the activated demonstrate either de novo or acquired resistance, particularly in CDK4/6:cyclin D1 complex removes the sequestration of E2F the metastatic setting (6, 7). Investigations into the mechanisms by Rb, allowing transcription of factors that promote progression underlying therapeutic resistance have focused on dysregulation of the cell cycle through the G –S-phase restriction point (17). of key signaling pathways downstream of ER and/or HER2 altera- 1 Mitogen-activated CDK4/6:Rb signaling can also occur inde- tions, including alterations in the PI3K/AKT/mTOR pathway and pendent of estradiol-mediated ER signaling. It is known that more recently the cyclin D:CDK4/6:Rb:p16 axis (8–12). activated HER2 signaling leads to increased transcription of cyclin D1 via activation of the PI3K and MAPK signaling pathways (18). Mitogenic control of the CDK4/6:Rb axis is frequently lost 1Division of Hematology/Oncology, Department of Medicine, Geffen School of through a series of molecular alterations known to occur to a 2 Medicine at UCLA, Los Angeles, California. Oncology Discovery Research, Lilly greater or lesser degree, in all subtypes of breast cancer (19, 20). Research Laboratories, Indianapolis, Indiana. These alterations are often complex and multifactorial, making it Note: Supplementary data for this article are available at Molecular Cancer difficult to predict which disease subtypes will respond to CDK4/6 Therapeutics Online (http://mct.aacrjournals.org/). inhibition in the absence of biologic data. Corresponding Authors: Dennis J. Slamon, Geffen School of Medicine at UCLA, Prior preclinical data from our laboratory demonstrated that þ 10945 Le Conte Avenue, 3360 PVUB, Los Angeles, CA 90095-3152. Phone: 310- a majority of luminal ER breast cancer cell lines are sensitive 825-5193; Fax: 1-310-825-6192; E-mail: [email protected]; and Neil both in vitro and in vivo to the CDK4/6 inhibitor, palbociclib, and O'Brien, E-mail: [email protected] that a combination of this drug with hormonal blockade is doi: 10.1158/1535-7163.MCT-17-0290 therapeutically synergistic in this subtype. These data led to the 2018 American Association for Cancer Research. clinical development and recent approval of palbociclib for use in

www.aacrjournals.org 897

O'Brien et al.

þ ER /HER2 metastatic breast cancer (13, 21–23) and subse- capture technology paired with Illumina HiSeq sequencing quently, a second CDK4/6 inhibitor, ribociclib in the same breast according with manufacturer's protocols (Supplementary Table cancer subtype (15). Abemaciclib (LY2835219; Eli Lilly) is also an S1). The deviations from consensus "normal" sequences in these orally available ATP-competitive inhibitor of CDK4 and CDK6, data were further filtered to enrich for mutations that are most with single-digit nanomolar potency against both kinases. Based likely to be somatic alterations with functional consequence in on early monotherapy clinical data (24) abemaciclib has been cancer using multiple published resources, most notably the given "Breakthrough Therapy" status from the FDA. This drug has COSMIC catalog of somatic mutations in cancer (http://www. now completed phase II and III clinical development in advanced sanger.ac.uk/science/tools/cosmic). DNA copy-number altera- þ ER /HER2 breast cancer resulting in its recent regulatory approv- tions were determined using comparative genomic hybridization þ al for ER /HER2 breast cancer either alone or in combination microarray assays (a-CGH) with the Agilent 105K oligonucleotide with hormonal blockade (25, 26). CGH chip according with manufacturer's protocols. Known onco- In the current study, we wanted to further evaluate the preclin- genes with log2 ratios >1 (2-fold) were considered amplified and ical activity of abemaciclib in cell lines representing each of the known tumor suppressor genes log2 ratios <0.8 were considered major therapeutic molecular subtypes of breast cancer and com- homozygous deletions. Baseline total and phosphoprotein levels pare it with other approved drugs in this class. Abemaciclib is of a list of >280 protein analytes enriched for proteins currently known to have increased selectivity for CDK4 over CDK6 as well known to be involved in cancer biology, were determined using as some inhibitory activity at other kinases such as CDK9, PIM1, reverse phase protein array (RPPA) from the core service at the MD HIPK2, and DYRK2 (27). In addition, we performed genomic and Anderson Cancer Center. Cell preparation and analysis were proteomic biomarker analyses in order to identify biomarkers performed in accordance with MD Anderson published protocols. and/or molecular profiles that correlate with sensitivity or resis- Data are presented here as log2 (intensity) values. Molecular tance to this molecule. Finally, we interrogated the potential for markers commonly tested in breast cancer, such as ER and HER2, therapeutically beneficial dual treatment regimens combining and several biomarkers that have previously been hypothesized to abemaciclib with cytotoxic therapies that are currently approved play a role in response to CDK4/6 inhibition were examined for and/or commonly used in the management of breast cancer. association with response to abemaciclib (listed in Supplemen- These data provide insight into biomarkers of response, other tary Table S2). molecular subtypes of breast cancer that may respond to CDK4/6 therapy and additional potential combination strategies that may In vivo efficacy studies þ þ þ be appropriate for clinical development. Xenograft models of ER /HER2 , HER2 /ER , and TNBC breast cancer cell lines were established in 6-week-old CD-1 athymic nude mice (Charles River Laboratories) as described in Materials and Methods þ þ Supplementary Materials. For ER /HER2 and HER2 /ER stud- Cell lines, cell culture, and reagents ies, 17-ß-estradiol 60-day release pellets (Innovative Research The growth inhibitory activity of abemaciclib (LY2835219- of America) were implanted subcutaneously into the left flank mesylate salt, provided by Richard Beckmann at Eli Lilly), palbo- 7 days before tumor inoculation. For in vivo studies, Fulvestrant ciclib (PD-0332991-HCL; Selleck Chemicals) and ribociclib (Faslodex, AstraZenica), trastuzumab (Herceptin, Genentech), (NVP-LEE011-succinate; MedChem Express) were assessed using and docetaxel (Taxotere, Hospira) were purchased from the UCLA a panel of 44 molecularly characterized human breast cancer cell pharmacy and 4-hydroxytamoxifen was purchased from Sigma- lines representing the known therapeutic subtypes of the disease Aldrich. Statistical differences between treatment arms at specific (22). Cells were cultured in appropriate culture media (e.g., time points were performed using a two-tailed paired Student t RPMI 1640, DMEM, L-15) supplemented with 10% to 15% test. Differences between groups were considered statistically heat-inactivated fetal bovine serum (FBS), 2 mmol/L glutamine, significant at P < 0.05. All statistics were calculated using Microsoft and 1% penicillin G-streptomycin-fungizone solution (PSF, Excel. All animal work was carried out under a protocol approved Irvine Scientific) as previously described (28). Cells were routinely by IACUC and the UCLA Animal Research Committee. assessed for mycoplasma contamination using a multiplex PCR fi GenePrint method and STR pro ling by the 10 System (Promega) siRNA knockdown of RB1 was used for cell-line authentication. Trastuzumab-resistant MDA-231 cells were seeded in 24-well plates at 10,000 cells per (BT-474-TR) cells were established as previously described (12, 29). well, followed by transfection with 1.25 pmol siRNA-targeting exon 19 of RB1 gene (Thermo Fisher Scientific, cat#4390824, In vitro proliferation assays s522), and 1.5 mL of lipofectamine (Thermo Fisher Scientific) in Cells were seeded into 24-well plates in the relevant culture media Opti-MEM I Reduced Serum Medium. Silencer Select Negative in duplicate at 5,000 to 50,000 cells per well, as described previously Control No.1 siRNA (Thermo Fisher Scientific, cat#4390843) was (22) and the following day 10 mmol/L of abemaciclib, palbociclib, used as a control in all assays. siRNA knockdown was confirmed or ribociclib with 2-fold serial dilutions over 6 to 12 concentrations by Western blot. was added to generate dose–response curves for IC50 determination (Supplementary Procedures). Drug combination studies using doc- Modaplex RT-PCR assay etaxel (Taxotere, Hospira) and Carboplatin (Hospira) were per- Total RNA was isolated from homogenized snap-frozen tumor formed as outlined in the Supplementary Procedures. tissue using the RNeasy kit from Qiagen. Specific cDNA was produced from the RNA using SuperScript VILO MasterMix as Molecular characterization of the breast cancer cell panel described by the User's Guide. mRNA expression of a set of cell- To identify potential predictive mutational biomarkers, whole cycle targets was quantified using the MODAplex RNA Array exome sequencing was performed using Agilent SureSelect hybrid multiplex qPCR platform (Qiagen). Data were exported from the

898 Mol Cancer Ther; 17(5) May 2018 Molecular Cancer Therapeutics

Preclinical Activity of Abemaciclib in Breast Cancer

A B HER2-/ER+ HER2+/ER- 10 EFM-19 ZR-75-1 MDA-361 MDA-468 HER2+/ER+ TNBC Abema 200 nmol/L -+ -+ -+ -+ 1 pRbS807/811 μmol/L

50 Total-Rb 0.1

Log IC Cyclin D1 0.01 CK-19

0.001 b-Acn BT-20 T-47D MCF-7 HCC70 HCC38 JIMT-1 CAL-51 BT-474 BT-549 Hs578T EFM-19 ZR-75-1 SK-BR-3 HCC202 DU4475 CAMA-1 ZR-75-30 HCC1143 HCC1806 HCC1954 HCC1395 HCC1419 HCC2218 HCC1569 HCC1187 HCC1500 HCC1937 COLO-824 UACC-893 UACC-812 UACC-732 EFM-192A SUM-190PT MDA-MB-231 MDA-MB-361 MDA-MB-436 MDA-MB-468 MDA-MB-157 MDA-MB-453 MDA-MB-415 SUM-225CWN MDA-MB-134-VI MDA-MB-175-VII C D E

1.00E+01 1.00E+01 1.00E+01

μmol/L R² = 0.7606 R² = 0.6447 μmol/L R² = 0.5956 μmol/L 50 1.00E+00 50 1.00E+00 1.00E+00 50 1.00E-02 1.00E-01 1.00E+00 1.00E+01 1.00E-02 1.00E-01 1.00E+00 1.00E+01 1.00E-02 1.00E-01 1.00E+00 1.00E+01

1.00E-01 1.00E-01 1.00E-01 Ribociclib Log IC Ribociclib Log IC Palbociclib Log IC Palbociclib 1.00E-02 1.00E-02 1.00E-02

Abemaciclib Log IC50 μmol/L Abemaciclib Log IC50 μmol/L Palbociclib Log IC50 μmol/L

Figure 1.

Activity of abemaciclib in breast cancer cell lines. A, In vitro IC50 values (generational inhibition) for each of the breast cell lines. Data, mean IC50 95% conﬁdence interval where available. B, Effect of single-dose (200 nmol/L) abemaciclib treatment on Rb signaling, cell lysates were prepared 5 days postdosing.

C, Scattergram comparing the in vitro IC50 values of abemaciclib with palbociclib in matching breast cancer cell lines; D, abemaciclib versus ribociclib; E, palbociclib versus ribociclib, for scattergrams, cell line subtypes are color coded consistent with A. R values were calculated in Excel. All experiments were repeated in at least duplicate.

MODAplex software and analyzed in Microsoft Excel. Mean abemaciclib at IC50 values below 500 nmol/L (Fig. 1A; Table 1). In expression values for each transcript were generated from at least general, sensitivity to abemaciclib was associated with high ER three replicate tumor samples. Expression values for each target levels, Rb wild-type status, high Rb total, and phosphoprotein transcript are represented as a ratio of the mean expression in the levels, lack of cyclin E amplification (normal copy number), low experimental arm relative to the vehicle control arm of the study. A cyclin E protein, p16 deletion, and/or low p16 protein levels "dose-dependent" response was defined as a >10% inhibition at (Supplementary Table S1). low dose (50 mg/kg) abemaciclib accompanied by a further In the same panel of cell lines, strong correlations were iden- increase in inhibition of >10% at higher dose (75 mg/kg) of tified between response to abemaciclib and two other CDK4/6 abemaciclib. inhibitors, palbociclib and ribociclib (Fig. 1C–E). Hormone receptor–positive cell lines (either HER2 normal or amplified) were commonly sensitive to all three molecules. In comparative Results analyses, abemaciclib was the most potent of the three molecules Activity of abemaciclib in the breast cancer cell line panel tested, followed by palbociclib and then ribociclib. In biomarker- The antiproliferative activity of abemaciclib was assessed in a positive breast cancer cell lines, the average (geometric mean) panel of 44 breast cancer cell lines, representing the known IC50 of abemaciclib was 168 nmol/L, compared with 306 nmol/L histological subtypes of the disease (see Supplementary Table for palbociclib and 913 nmol/L for ribociclib. S2 for full details). The cell lines were differentially responsive to abemaciclib over a wide concentration range, with IC50 values Abemaciclib in combination with hormone blockade in þ varying between 0.012 and 3.73 mmol/L (Fig. 1A). On-mecha- ER /HER2 breast cancer cell xenografts nism activity of abemaciclib was confirmed by a loss of Rb Combined activity of abemaciclib and hormone blockade was þ phosphorylation in response to treatment (Fig. 1B) followed by confirmed in two ER breast cancer cell lines xenografts (Fig. 2 and induction of G1 cell-cycle arrest (Supplementary Fig. S1). No loss Supplementary Figure S2). Significant tumor regressions were þ of pRb or induction of G1 cell-cycle arrest was observed in the Rb- observed with single-agent abemaciclib in MCF-7 (ER /HER2 ) deficient MDA-468 cell line (Fig. 1B; Supplementary Fig. S1). xenografts, and combination with either tamoxifen or fulvestrant þ ER /HER2 and HER2-amplified breast cancer cell lines that induced a marginal increased benefit in antitumor response þ þ express high levels (>median) of ER protein (i.e., HER2 /ER ; (Fig. 2A and B; Supplementary Table S3). However, within the Fig. 1A; Supplementary Table S2) were among the most sensi- 1st week of drug withdrawal, tumors began to progress in the tive to abemaciclib. In addition, despite relatively low levels of mice treated with 50 mg/kg of abemaciclib alone as well in ER and HER2, a subset of TNBC cell lines also responded to the mice treated with either single-agent tamoxifen or fulvestrant.

www.aacrjournals.org Mol Cancer Ther; 17(5) May 2018 899

O'Brien et al.

Table 1. Biomarkers of response to abemaciclib triple-negative breast cancer cell lines

IC50 PIK3CA RB1 RB1 CDKN2A CDKN2A CCNE1 Cell line mmol/L ER HER2 PMs PMs CN Rb RbpS807S811 PMs CN p16 CN CyclinE1 HCC1395 0.310 -2.16 -0.1 Wt Wt GAIN 0.44 3.15 Wt HD -3.01 GAIN 0.27 BT-20 0.328 -1.61 1.52 p.Pro539Arg Wt NC 0.56 2.98 Wt HD -2.81 NC -0.33 CAL-51 0.407 -2.1 1.62 p.Glu542Lys Wt NC 0.89 3.84 Wt NC -2.15 NC 0.07 MDA-MB-231 0.443 -2.26 0.79 Wt Wt NC 0.65 3.43 Wt HD -2.73 GAIN -0.17 HCC1143 0.528 -1.15 0.97 Wt Wt NC 0.36 2.87 Wt LOH -2.55 NC -0.17 COLO-824 0.903 -1.7 1.56 Wt Wt HD -0.31 1.15 Wt GAIN -0.3 NC 0.48 HCC1806 0.904 -1.76 0.68 Wt Wt NC 0.35 3.07 Wt HD -2.96 HIGH AMP 1.98 Hs578T 0.968 -2.18 0.86 Wt Wt NC 0.74 3.56 Wt NC -1.39 NC -0.31 HCC70 0.971 -1.96 1.74 Wt Wt NC -0.08 0.8 Wt AMP -0.31 GAIN 0.21 MDA-MB-436 1.050 -1.98 0.66 Wt Wt GAIN -0.26 0.24 Wt NC -0.69 NC 0.83 MDA-MB-468 1.090 -1.95 -0.53 Wt Wt HD -0.41 1.52 Wt NC -1.02 AMP 0.27 HCC1937 1.180 -1.22 1.08 Wt c.2221*>-G LOH -0.32 1.13 Wt NC -0.51 NC 0.09 MDA-MB-157 1.520 -2.1 0.63 Wt Wt LOH 0.11 1.87 Wt NC -0.23 AMP 1.02 HCC38 2.280 -2.47 2.42 Wt Wt NC 1.48 3.78 Wt NC -3.29 NC -0.58 HCC1187 2.300 -1.92 2.37 Wt Wt NC -0.28 1.37 Wt NC -0.03 GAIN 1.09 DU4475 3.290 -2.17 -0.4 Wt Wt HD 0.28 3.4 Wt NC -2.21 NC 1.54 BT-549 3.730 -2.36 0.91 Wt Wt HD -0.54 0.43 Wt GAIN 0.64 NC 0.23 Abbreviations: CN, copy number; PM, point mutation.

Conversely, sustained inhibition of tumor growth was observed in response to the combination versus single-agent treatments for over 6 weeks after drug withdrawal in the majority of animals (Fig. 2D). This was accompanied by a reduction in mitosis as treated with either drug combination or with high-dose abema- indicated by a decrease in FOXM1 protein levels (Fig. 2D). ciclib monotherapy at 75 mg/kg (Fig. 2C). Molecular analysis of Western blot analysis of residual xenograft tissue collected at the xenograft tissue at day 4 of dosing revealed a greater loss of pRb end of a recovery phase of 6 to 7 weeks revealed that active cell

A B 300 Abe Abe F 1,200 MCF-7: ER+ BCL Vehicle Abe Fulv Tam Abe DAY 6: Vehicle Abema Fulvestrant 50mg/kg 75mg/kg + + 200x ) 250 Fulv Tam 3 1,000 200 800 * 150 600 100

400 50 Tamoxifen Abema+Fulvestrant Abema+Tamoxifen

200 0

Tumor volumeTumor (mm -50

0 Regressions (%) and ΔT/C 0 5 10 15 20 25 30 35 -100 Vehicle Abema 50 mg/kg Abema 75 mg/kg Progressive disease Stable disease Paral response Fulvestrant Abema+Fulv Tamoxifen -150 Complete response Abema+Tam C ) 1,600 Abema 50 mg/kg Abema 75 mg/kg 1,600 Tamoxifen 1,600 3 1,600 Fulvestrant 1,600 Abema + Tamox 1,600 Abema + Fulv

1,200 1,200 1,200 1,200 1,200 1,200

800 800 800 800 800 800

400 400 400 400 400 400

Tumor volumeTumor (mm 0 0 0 0 0 0 Days 0 153045607590 0 153045607590 0 153045607590 0 153045607590 0153045607590 0 153045607590 Treatment withdrawn Treatment withdrawn Treatment withdrawn Treatment withdrawn Treatment withdrawn Treatment withdrawn

D On treatment (Day 3) E > 6 weeks post drug withdrawal Vehicle Abema Fulvestrant Abe + fulv Tamox Abe + tam Vehicle Abema Fulvestrant Abema + fulv Tamox Abema + tam 1R 1B 2L 2R 4N 4B 4N 5N 5R 6B 6N 7N 7B 1N 1R 1L 1B 2L 2N 2B 4R 4B 4N 5R 5B 5L 6L 6R 6N 7R 7L 7N pRbS807/811 pRbS807/811 Total-Rb Total-Rb FOXM1 Cyclin-D FOXM1 Keran-19 b-Acn b-Acn

Figure 2. Induction of sustained tumor growth inhibition by abemaciclib plus hormone blockade. A, Growth curves for MCF-7 (ERþ) breast cancer cell line (BCL) xenografts treated with either single abemaciclib or combination with hormonal blockade (fulvestrant or tamoxifen), 8 mice per arm. For combination arms, 50 mg/kg abemaciclib was used. Data, mean tumor volume SEM. B, Waterfall plots representing the % progression or tumor regression in each individual tumor within each treatment arm, 20% progression was considered stable disease, 50% tumor regression was considered a partial response, and 100% regression a complete response. Fulv, Fulvestrant; Tam, Tamoxifen. C, Growth curves for individual xenograft tumors posttreatment withdrawal relative to mean tumor volumes for the vehicle control. D, Western blot analysis of snap-frozen tumor tissue taken from mice 24 hours after the 3 days of continuous dosing. E, Western blot of representative tumor lysates taken >6 weeks postdrug withdrawal. F, Representative images of cells stained for b-galactosidase 6 days posttreatment with the molecules indicated. , Statistically signiﬁcant difference from vehicle control; P < 0.05.

900 Mol Cancer Ther; 17(5) May 2018 Molecular Cancer Therapeutics

Preclinical Activity of Abemaciclib in Breast Cancer

division and Rb signaling were restored in the 50 mg/kg abema- and tumor regressions in xenografts progressing on trastuzumab ciclib and single-agent endocrine treatments, as indicated by the alone (Fig. 3B and C; Supplementary Table S4). The triple com- recovery of FOXM1 and pRb protein levels (Fig. 2E). In contrast, bination of abemaciclib, trastuzumab, and tamoxifen further FOXM1/pRb signal remained low in xenograft tissues collected improved tumor regressions in both models (Fig. 3A–C; Supple- from combination arms at the same time point (Fig. 2E). To mentary Table S4). In vitro studies confirmed that triple blockade investigate if the sustained reduction in Rb/FOXM1 signal in the of CDK4/6, HER2, and ER signaling leads to a greater induction combination arms was due to induction of permanent growth of cell death in both trastuzumab-sensitive and -resistant þ þ arrest/senescence or induction of cell death, we measured levels of HER2 /ER breast cancer cell lines (Fig. 3D). Abemaciclib was the human epithelial marker protein keratin-19 (CK-19), a struc- also shown to combine effectively with docetaxel in these HER2- tural protein not directly regulated by abemaciclib treatment amplified breast cancer cell line xenografts. The addition of (Supplementary Fig. S3). In the MCF-7 xenograft tissues, CK-19 docetaxel to abemaciclib did not inhibit the TGI induced by protein levels were significantly reduced or lost in the combination abemaciclib or docetaxel alone. Moreover, the combination of treatment arms, particularly in abemaciclib plus tamoxifen treated abemaciclib, trastuzumab, tamoxifen, and docetaxel was the mice, indicating loss of human epithelial tumor cells (Fig. 2E). most efficacious arm in both studies (Fig. 3A–C; Supplementary Using a ß-galactosidase–staining assay, in vitro assays of the MCF-7 Table S4). Marginal and reversible body weight loss was observ- cells showed increased induction of senescence in response to the ed in the mice treated with docetaxel containing arms over the combination as opposed to the single agent (Fig. 2F). 4 weeks treatment period (Supplementary Table S4). Western blot analysis of xenograft tissues collected after 4 days þ þ Activity of abemaciclib in HER2 /ER breast cancer cell lines of treatment confirmed a dose-dependent loss of pRb, TOPOIIa, and xenografts and pHH3 in mice treated with abemaciclib (Fig. 3E). Further Activity of abemaciclib single agent or combination with HER2 reduction in Rb signaling was observed in xenografts treated with and ER-targeted therapy was assessed in two xenograft models of abemaciclib plus trastuzumab or trastuzumab plus tamoxifen. þ þ HER2 /ER breast cancer, parental BT474 cells and BT474 cells The addition of docetaxel did not block the pRb knockdown conditioned through long-term drug exposure to progress on induced by abemaciclib or the combination of abemaciclib with trastuzumab therapy (BT-474-TR (12)). Single-agent abemaciclib trastuzumab plus tamoxifen (Fig. 3E). induced significant tumor growth inhibition (TGI) in both models (BT-474; P ¼ 0.028, BT-474-TR; P < 0.001; Fig. 3A–C; Sup- Identification of a subset of TNBC cells sensitive to abemaciclib plementary Table S4). The combination of abemaciclib with In TNBC cell lines, where cell growth is independent of ER trastuzumab induced significantly improved (P ¼ 0.0012) TGIs status, response to abemaciclib appears to remain dependent on

A BT-474, HER2+/ER+, Trastuzumab sensive B BT-474-TR, HER2+/ER+, Trastuzumab acquired resistant C 600 1,000 550

) BT-474 ) 3

3 450

500 ) 3 800 BT-474-TR 400 * 350 600 300 * 250

# 400 # 200 150

Tumor volume (mm 200 100 Tumor volume (mm 50 Δ volumeTumor (mm in 0 0 -50 0 4 8 1216202428 0481216202428 Days Days Vehicle Abe 50 mg/kg Trastuzumab Vehicle Abe 50 mg/kg Abe 75 mg/kg -150 Abe+Tz Abe+Dx Abe+Tz+Tam Trastuzumab Abe+Tz Abe+Dx Abe+Tz+Dx+Tam Abe+Tz+Tam Abe+Tz+Dx+Tam D E 35 Vehicle Abe (50) Abe (75) Trast (Tz) Abe (50)+Tz Abe+Dtx Abe+Tz+Tam Abe+Tz+Tam+Dtx 30 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

cells 25

+ pRbS807/811 20 TOPO IIα 15 PHH3 10 Total HER2 % Annexin V 5 0 b-Acn BT474 BT-474-TR Vehicle Abema 200 nmol/L Traz 15 µg/mL Tam 500 nmol/L Traz+Tam Abema+Tam Abema+Traz Abema+Traz+Tam

Figure 3. Activity of abemaciclib in combination with HER2 and ER-targeted therapies in HER2þ/ERþ breast cancer. A and B, Growth curves for BT-474 and trastuzumab conditioned BT-474-TR cell line xenografts treated with either single agent abemaciclib ("Abe") or combination with HER2-targeting (trastuzumab, "Tz"), ER-targeting (tamoxifen, "tam") or SOC chemotherapy (docetaxel, "Dtx"), 8 mice per arm. Fifty mg/kg abemaciclib was used in combination arms. C, Change in tumor volume from baseline for each experimental arm. D, Induction of apoptosis in cell line in 2D cultures treated with combination CDK4/6, HER2/ER combination therapy. Histograms, % annexin V-positive (AN Vþ) cells after 5 days of treatment. Data, mean SD. All experiments were repeated in at least duplicate. E, Lysates from snap-frozen BT474-TR tumor tissue, collected after 4 days of dosing, were subject to immunoblotting with the antibodies indicated. , Statistically signiﬁcant difference compared with vehicle control; #, statistically signiﬁcant difference compared with single-agent trastuzumab; P < 0.05.

www.aacrjournals.org Mol Cancer Ther; 17(5) May 2018 901

O'Brien et al.

Vehicle Abema 50 Abema 75 C HCC70 MDA231 A MDA-231, High pRb, Low p16 MDA-231: Day 21 500 200 1N 1R 1L 1N 1L 1N 2L 2R 3L 3N Vehicle Abe Abe Docetaxel Abema S807/811 ) 50 mg/kg 75 mg/kg + S807/811 pRb

3 pRb 400 150 Dtxl p16 Total-Rb 300 100 b-Acn FOXM1 200 50 Cyclin D 100 0 b-Acn Tumor volumeTumor (mm 0 0 3 6 9 12 15 18 21 Regressions (%) and ΔT/C -50 Vehicle control Abemaciclib 50 mg/kg QD D Abemaciclib 75 mg/kg QD Docetaxel 10 mg/kg QW -100 100 Progressive disease Stable disease Paral response Abemaciclib + Docetaxel 90 Untransfected Complete response 76.4 75.8 80 Scrambled control 70 Rb1 Knockdown B 60 46.4 49.4 150 1,200 HCC70, Low pRb, High p16 Vehicle Abe Abe Docetaxel Abema 50 43.5 + 50 mg/kg 75 mg/kg 40 ) Dtxl 23.7 3 1,000 23.9 100 30 20.6 15.0 800 20 10

50 inhibion % Generaonal 600 0 1,000 nmol/L Abema 200 nmol/L Abema 40 nmol/L Abema 400 0 Day 0 Day 2 Day 5 200 Tumor volumeTumor (mm

0 (%) Regressions and ΔT/C -50 0 3 6 9 12 15 18 21 Abema 200 nmol/L Vehicle control Abemaciclib 50 mg/kg QD ------+ + + - - - + + + -100 S807/811 Abemaciclib 75 mg/kg QD Docetaxel 10 mg/kg QW Progressive disease Paral response pRb Abemaciclib + Docetaxel tRb

b-Acn

Figure 4. Abemaciclib has activity in xenograft models of TNBC. A and B, Growth curves for TNBC cell line xenografts treated with either single abemaciclib ("Abe") or combination with SOC chemotherapy (docetaxel "dtxl"), 8 mice per arm. Fifty mg/kg abemaciclib was used in combination arms. C, Left, Western blot analysis of snap-frozen tumor tissue taken from baseline vehicle control arms of each xenograft study. Right, Western blot analysis of snap-frozen tumor tissue, collected from mice in each arm of the MDA-231 study, 24 hours after the 3 days of continuous dosing. D, siRNA knockdown of Rb1 confirms Rb1 signaling is required for abemaciclib response. Histogram, % generational inhibition in response to 5 days abemaciclib treatment in MDA-231 with and without Rb1 siRNA knockdown. Western blots confirm knockdown of Rb1 by siRNA at the time points indicated. #, Statistically significant versus scrambled nontargeting control, P value according to Student t test.

intact Rb signaling. Those TNBC cell lines that have high Combined activity of abemaciclib and cytotoxic baseline levels of total and phosphorylated pRb, accompanied chemotherapy by low levels of p16 protein, are among the most sensitive to The potential of abemaciclib to be used in combination with abemaciclib (Table 1). Cell lines with copy-number gains or cytotoxic chemotherapy was further investigated in a cell line amplification of the CCNE1 (cyclin E1) gene are clearly less model of TNBC in vitro. In order to determine if the lack of responsive (Table 1). antagonism observed in our in vivo models could be attributed Xenografts of TNBC cell lines expressing sensitive and resis- to an effect of the timing of drug administration, we investigated tant biomarker profiles were measured for response to abema- dosing strategies comparing coadministration versus sequencing ciclib single agent and combination with docetaxel, a standard- of these treatments. of-care agent used for TNBC. Consistent with all the preceding Simultaneous treatment of MDA-231 cells with abemaciclib data, induction of tumor regressions or stable disease was plus docetaxel or carboplatin resulted in increased inhibition observed only in the MDA-231 (Fig. 4A; Supplementary Table of cell proliferation compared with single-agent treatments S5) and BT-20 xenografts (Supplementary Fig. S4) that have (Fig. 5A). Concentrations of abemaciclib below 10 nmol/L high levels of pRb and low levels of p16 at baseline (Fig. 4C. induced profound (>60%) inhibition of cell proliferation when left). Conversely, HCC70 xenografts with low levels of pRb and combined with either antimitotic agent. Pretreatment of cells high p16, continued to progress through abemaciclib mono- with abemaciclib for 2 days reduced cell proliferation rate via therapy (Fig. 4B). On-target activity of abemaciclib was con- induction of G1 cell-cycle arrest (Supplementary Fig. S5A) firmed in the sensitive MDA-231 xenografts by reduction in without impacting the activity of docetaxel or carboplatin total and phosphorylated Rb and the cell-cycle progression relative to cells pretreated with vehicle control (Fig. 5B and marker, FOXM1, in response to treatment (Fig. 4C, right). C). Pretreatment with abemaciclib for 24 hours also induced þ þ Consistent with our observations in HER2 /ER breast cancer G1 cell-cycle arrest (Supplement Fig. S5B) without blocking xenografts, combination of abemaciclib with docetaxel did not apoptosis induced by high-dose docetaxel treatment. In con- antagonize the activity of either single agent (Fig. 4A and B). trast, a dose-dependent decrease in apoptosis induction by Finally, the role of pRb in predicting response to abemaciclib in docetaxel was observed in response to pretreatment with palbo- TNBC was confirmed by siRNA knockdown of RB1 gene expres- ciclib or ribociclib (Fig. 5D). Coadministration of docetaxel sion in the MDA-231 cells. Subsequent reduction of baseline with either palbociclib or ribociclib after CDK4/6-inhibitor pRb levels significantly reduced the sensitivity of the MDA-231 pretreatment blocked apoptosis induction, whereas cotreat- cells to abemaciclib (Fig. 4D). ment with abemaciclib did not (Fig. 5E).

902 Mol Cancer Ther; 17(5) May 2018 Molecular Cancer Therapeutics

Preclinical Activity of Abemaciclib in Breast Cancer

A Abemaciclib (Abe) B 120 12,000 C 12,000 Docetaxel (Dtx) Growth Day 2 to Day 5 Growth Day 2 to Day 5 Carboplan (Carb) 100 10,000 Growth Day 0toDay 2 10,000 Abe+Dtx Growth Day 0 to Day 2 80 Abe+Carb 8,000 8,000

* 6,000 6,000 60 * *

4,000 Cell number

Cell number 4,000 40 2,000 2,000 20 % Generaonal inhibion 0 0 0 Abema (nmol/L) 1 0.2 0.04 0.008 0.0016 0.00032 Dtx (nmol/L) 10 2.0 0.40 0.080 0.0160 0.00320 Veh/Veh Veh/Abe Abe/Veh Veh/Dtx Abe/Abe Abe/Dtx Veh/Veh Veh/Abe Abe/Veh Carb ( mol/L) 25 12.5 6.25 3.125 1.5635 0.78125 Abe/Abe Veh/Carb Abe/Carb Veh/Abe+Dtx Abe/Abe+Dtx Veh/Abe+Carb Abe/Abe+Carb D E Vehicle/Vehicle Vehicle/Dtx 100 nmol/L 20 18 18 16 16 14 14 # # cells cells # 12 # 12 + + 10 10 # # 8 Ribo/Dtx 100 nmol/L 8 # Abema/Dtx 100 nmol/L Palbo/Dtx 100 nmol/L # # 6 6 4 4 Propidium iodide

2 % Annexin V 2 % Annexin V 0 0 ib+Dtx ib+Dtx ib+Dtx Veh/Veh Veh/Veh

Annexin VFITC /Abe+Dtx /Abe+Dtx nmol/L,R nmol/L,R nmol/L,R nmol/L/Plb+Dtx nmol/L/Plb+Dtx nmol/L/Plb+Dtx nmol/L/Abe+Dtx nmol/L nmol/L Plb 200 nmol/L/Dtx Plb 500 nmol/L/Dtx Rib 200 nmol/L, Dtx Rib 500 nmol/L, Dtx Veh/Dtx100 nmol/L Veh/Dtx 100 nmol/L Abe 200 nmol/L/Dtx Abe 500 nmol/L/Dtx Plb 1,000 nmol/L/Dtx Rib 1,000 nmol/L, Dtx Abe 1,000 nmol/L/Dtx Rib 200 Rib 500 Plb 200 Plb 500 Rib 1,000 Abe 500 Abe 200 Plb 1,000 Abe 1,000

Figure 5. Abemaciclib combined with cytotoxic chemotherapy induces increased inhibition of cell proliferation and induction of cell death. A, MDA-231 TNBC cells were treated simultaneously with a range of concentrations of abemaciclib plus docetaxel or carboplatin for 5 days. B and C, Cells were pretreated for 48 hours with or without 200 nmol/L abemaciclib followed by treatment with 1 nmol/L docetaxel (dtx) or 25 mmol/L carboplatin (carb) or the combination of abemaciclib plus docetaxel or carboplatin for the remaining 72 hours. Total cell counts were measured at day 0, day 2 (washout), and day 5. D, Annexin V staining of cells pretreated with or without increasing concentrations of abemaciclib, palbociclib, or ribociclib for 24 hours followed by 24 hours of treatment with 100 nmol/L docetaxel, left panels show representative examples, summary histogram on the right. E, Annexin V staining of cells pretreated for 24 hours with a range of concentrations of either abemaciclib, palbociclib, or ribociclib followed by switch to 100 nmol/L docetaxel plus abemaciclib, palbociclib, or ribociclib. Veh, vehicle control/untreated. Data, mean SD. All experiments were repeated in at least duplicate. , Statistically signiﬁcant difference compared with vehicle control; #, statistically signiﬁcant difference compared with vehicle/docetaxel arm; <0.05, P values according to Student t test.

þ þ þ Identification of a pharmacodynamic signature of response to (ER /HER2 ), and BT-474-TR (HER2 /ER ) xenografts abemaciclib treatment (Fig. 6D). Expression of five of these transcripts, RRM2, TOPO2A, To identify potential pharmacodynamic (PD) markers of MKI67, MCM7, and CDK2 are directly regulated by the E2F- response to abemaciclib, we used a Modaplex PCR-based plat- transcription factor immediately downstream of Rb, indicating form to measure changes in the expression of a set of 22 cell-cycle– a direct on-target regulation of these genes by CDK4/6 inhibition associated genes in pre- and posttreatment (day 4) samples from using abemaciclib. tumor tissues responding to abemaciclib-based therapy. In MCF-7 þ ER /HER2 xenografts, a range of mRNA expression changes were induced by treatment with abemaciclib, hormonal blockade or a Discussion combination of the two (Fig. 6A). A subset of 10 transcripts The cyclinD:CDK4/6:Rb:p16 signaling axis is frequently dysre- showed a dose-dependent reduction in expression in response gulated in cancer. Efforts to pharmacologically target this pathway to low- and high-dose abemaciclib. Single-agent treatment with have been validated by the approval of the specific CDK4/6 either fulvestrant or tamoxifen had little effect on the expression of inhibitor, palbociclib (PD-0332991, Pfizer), ribociclib (LEE-011, these 10 candidate PD biomarkers. However, combination of Novartis), and most recently abemaciclib (Eli Lilly), in combi- þ low-dose abemaciclib (50 mg/kg) with either fulvestrant or nation with endocrine-based therapies in advanced ER /HER2 tamoxifen increased the inhibition of the candidate gene set breast cancer (13, 15, 21–23, 25, 26). Abemaciclib is structurally compared with inhibition induced by high-dose abemaciclib and biologically distinct from palbociclib and ribociclib with (75 mg/kg; Fig. 6B). Using the same criteria, an abemaciclib greater selectivity for CDK4 over CDK6 (27) and clinically, þ þ dose-dependent gene signature was identified from ER /HER2 abemaciclib appears to have superior single-agent activity when þ xenografts. Similar to the ER /HER2 models, expression of these compared with the other approved CDK4/6 inhibitors (24, 25, transcripts was similarly regulated by CDK4/6-inhibition in com- 30, 31). It is possible that the CDK4 selectivity of abemaciclib bination with either trastuzumab or tamoxifen or both (Fig. 6C). contributes to the improvement in activity by reducing the degree Transcripts for RRM2, TOPO2A, MKI67, MCM7, CDK2, and CDK4 of neutropenia usually associated with inhibition of CDK6, were all found to be regulated by abemaciclib in a dose-dependent which in turn avoids the need for an interruption in dosing þ manner in sensitive tumors from MCF-7 (ER /HER2 ), ZR-75-1 associated with both palbociclib and ribociclib. This difference

www.aacrjournals.org Mol Cancer Ther; 17(5) May 2018 903

O'Brien et al.

A MCF-7 (ER+) Xenogra s B MCF-7 (ER+) Xenogra s 50 mg/kg Abemaciclib 75 mg/kg Abemaciclib CCNB1 RRM2 TOPO2A MKI67 E2F4 E2F1 CDKN1A MCM7 100 CDK4 PTEN E2F2 CDK2 RB-1 EX17-18 MYC Cyclin E E2F3 Abemaciclib (50 mg/kg) + fulvestrant Abemaciclib (50 mg/kg) + Tamox RBL1 CCND3 RB-1 EX25-26 CCND1 E2F5 RBL2 90 100 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 % Inhibion of gene expression of % Inhibiongene % Inhibion of gene expression of % Inhibiongene 50 mg/kg Abemaciclib 75 mg/kg Abemaciclib Fulvestrant Abemaciclib (50 mg/kg) Tamoxifen Abemaciclib (50 mg/kg) CCNB1 RRM2 TOPO2A MKI67 E2F1 MCM7 CDK4 PTEN E2F2 CDK2 + fulvestrant + Tamox C BT474-TR (HER2+/ER+) Xenogra s D 100 50 mpk Abemaciclib MCF-7 ZR-75-1 BT474-TR 90 75 mpk Abemaciclib 50 mg/kg 75 mg/kg 50 mg/kg 75 mg/kg 50 mg/kg 75 mg/kg Abemaciclib Abemaciclib Abemaciclib Abemaciclib Abemaciclib Abemaciclib 50-AB + Trastuzumab 80 CCNB1 74.52 92.05 0.50 13.45 88.73 94.00 50-AB + Tamox RRM2 70 67.20 96.56 52.28 91.20 31.53 93.37 50-AB + Trastuzumab + Tamox TOPO2A 61.15 84.21 76.94 94.43 78.73 94.33 60 MKI67 55.43 88.23 70.66 94.22 74.50 94.50 50 E2F1 32.82 84.09 0.00 55.28 0.00 65.70 40 MCM7 29.38 76.18 25.37 74.84 28.30 77.37 CDK4 30 27.21 38.17 17.10 35.95 26.87 44.57 PTEN 15.86 57.55 31.22 30.10 0.00 0.00 20 E2F2 14.04 84.61 18.53 64.21 -1.93 14.77 10 % Inhibion of gene expression of % Inhibiongene CDK2 11.69 46.86 20.32 57.23 11.03 64.63 0 RBL1 0.00 13.25 24.16 52.63 1.13 0.00 TOPO2A MKI67 FOXM1 RRM2 MCM7 CDK4 RB-1 EX25-26 RB-1 EX17-18 CDK2

Figure 6. Identiﬁcation of PD biomarkers of abemaciclib response in vivo. A, Percent inhibition of mRNA expression of 22 cell-cycle–regulated genes, relative to vehicle control, in each of the treatment arms in tumor collected snap-frozen 24 hours after the 3 days of continuous dosing in the MCF-7 xenograft study. B, Percent inhibition of a subset of genes that show abemaciclib dose dependence and response to abemaciclib plus hormone blockade therapy. C, Abemaciclib dose- dependent genes in HER2þ/ERþ breast cancer cell line xenografts are also responsive to HER2/ER-targeted therapy. D, Subset of genes commonly altered by abemaciclib treatment across three breast cancer cell line xenograft studies. Values, Mean percent inhibition of gene expression relative to vehicle control in each study.

may improve the chances of forcing tumor cells into permanent Our studies also confirm that abemaciclib has additive and/or growth arrest and ultimately senescence (24). In this study, we synergistic activity in combination with endocrine-based thera- þ present a comprehensive analysis of the preclinical activity of pies in cell line xenograft models of ER /HER2 breast cancer. abemaciclib using both in vitro and in vivo preclinical models Continuous daily treatment using clinically achievable doses of spanning the known molecular subtypes of breast cancer. Using abemaciclib (50 mg/kg) in combination with tamoxifen or ful- pre- and posttreatment data, we have identified a set of consistent vestrant was well tolerated and led to downregulation of Rb þ þ biomarkers of sensitivity to abemaciclib in ER , HER2 , and signaling and significantly improved antitumor responses com- TNBC subtypes. pared with results with either agent alone. While tumors from Measurement of the growth inhibitory activity of abemaci- single-agent–treated animals began to proliferate within days of clib across a panel of 44 human breast cancer cell lines iden- treatment cessation, the majority of xenografts from mice treated þ tified the ER /HER2 subtype as most sensitive to CDK4/6 with combination therapy showed no signs of tumor progression, inhibition. Despite the wide therapeutic range of this molecule, despite withdrawal of treatment for more than 6 weeks. Xeno- þ each of the 9 cell lines classified as ER /HER2 had IC50 values grafts collected during the dosing phase of the combination below 200 nmol/L, making further stratification of response studies indicate an initial induction of cellular senescence as biomarkers within this subtype unnecessary. This activity was measured by loss of FOXM1 and in vitro assays showing a greater þ confirmed in two xenograft models of ER /HER2 breast induction of senescence with the combination treatment. Xeno- cancer by the induction of complete arrest of tumor growth graft materials collected at the end of the study indicate a loss of using abemaciclib monotherapy. Inhibition of tumor prolifer- human epithelial tumor cell population. The marked tumor ation was accompanied by dose-dependent decreases of phos- regressions observed in the combination arms provide further phorylated Rb and induction of cell-cycle arrest as measured by evidence of induction of tumor cell death induced by this regi- loss of the cell-cycle progression markers, TOPOIIa (S-phase), men. The data reported here are consistent with those recently phosphohistone-H3 (G2–M), and FOXM1 (cellular senes- reported from the MONARCH-2 clinical trial (NCT02107703), cence). At the tested doses, abemaciclib did not inhibit which showed significant improvements in progression-free sur- CDK-9 signaling, consistent with previous data reported for vival (PFS) and objective response rates (ORR) in patients with this molecule (27, 32). Comparison of the activity of abema- advanced breast cancer treated with abemaciclib plus fulvestrant ciclib, palbociclib, and ribociclib across the panel of 44 breast versus fulvestrant alone (26). cancer cell lines identified a strong correlation within and Abemaciclib-sensitive cell lines were also identified within the between the major histologic subtypes of breast cancer for this HER2-amplified subgroup. There is considerable preclinical evi- class of molecule. However, of the three, abemaciclib appeared dence to support targeting CDK4/6 in HER2-amplified breast to be the most potent using our assays. cancers given that HER2 signaling can drive cell-cycle progression

904 Mol Cancer Ther; 17(5) May 2018 Molecular Cancer Therapeutics

Preclinical Activity of Abemaciclib in Breast Cancer

through activation of cyclinD1:CDK4/6 signaling (18, 33). Cyclin greater inhibition of tumor cell growth compared with single- D1 is required for the formation of HER2-initiated tumors in agent treatment. The combined activity observed between these mouse models and pharmacologically targeting CDK4/6 blocks two classes of molecules could possibly be explained by fact that tumor formation in mice (34, 35). In addition, preclinical studies not all tumor cells will be uniformly arrested by abemaciclib show that targeting CDK4/6 overcomes resistance to HER2-direct- leaving subpopulations of cells sensitive to the mechanism of ed therapy both in vitro and in vivo (9, 36). The data presented here action of antimitotics. It has previously been shown that abema- provide further insight into the specific subtypes of HER2-ampli- ciclib can be combined with gemcitabine without antagonism in fied breast cancers most likely to benefit from CDK4/6-targeted lung cancer xenografts with either sequential or simultaneous therapy. Analysis of abemaciclib response within the panel of 17 administration (27). Our in vitro experiments also show that HER2-amplified breast cancer cell lines indicates that HER2- abemaciclib appears to be unique among the CDK4/6 inhibitors amplified cell lines with higher levels of ER protein accompanied tested. In contrast to abemaciclib, both palbociclib and ribociclib by intact downstream Rb signaling (high total Rb and pRb and no pretreatment significantly reduced the induction of apoptosis cyclin E amplification) are most sensitive to abemaciclib treat- by docetaxel in a dose-dependent manner. The specific mechan- ment. These data indicate that HER2-amplified breast cancer with isms by which abemaciclib differs from palbociclib and ribociclib þ þ higher levels of ER (i.e., HER2 /ER ) may be more susceptible to in this setting requires further investigation; however, these þ CDK4/6 intervention than those that are HER2 /ER . The current data are encouraging for the development of abemaciclib in data demonstrate that single-agent abemaciclib induces signifi- indications where combination with cytotoxic chemotherapies cant tumor growth inhibition as well as dose-dependent inhibi- might be required. þ þ tion of pRb signaling and cell-cycle arrest in HER2 /ER xeno- Sensitivity to abemaciclib was also detected in a specific sub- grafts. In addition, abemaciclib combined effectively with trastu- group of TNBC cell lines matching the biomarker profile of intact zumab and in triple combination with tamoxifen, induced sig- Rb signaling, despite expressing low levels of ER and HER2. Of the nificant regressions in both trastuzumab-sensitive and -resistant 17 TNBC cell lines tested, the top 5 most sensitive cell lines had þ þ HER2 /ER xenografts. In vitro studies confirmed that the triple high levels of total and phosphorylated levels of Rb accompanied combination of targeting CDK4/6, HER2, and ER leads to a greater by low baseline levels of p16. Abemaciclib IC50 values for this induction of cell death in cells when compared with single-agent subset of TNBC cell lines are similar to those measured for the þ þ þ treatment. The specific mechanism by which this triple combi- ER /HER2 and HER2 /ER cell lines. Molecular alterations nation is effective remains to be determined. Finally, abemaciclib associated with resistance to CDK4/6-targeted therapy (43, 44) activity in HER2-amplified tumors progressing on trastuzumab through activation of Rb signaling independently of CDK4/6, provides encouragement that targeting CDK4/6 may be effica- such as high p16 protein levels or amplification of cyclin E, were cious in treatment refractory HER2-amplified disease. These data found in those TNBC cell lines least sensitive to abemaciclib. also support the hypothesis that ER and downstream Rb signaling Xenograft studies confirmed that TNBC cell lines selected for high þ þ are, at least partially, driving the progression of HER2 /ER baseline levels of pRb and low p16 were sensitive to abemaciclib þ þ þ tumors. Clinical data suggest that HER2 /ER and HER2 /ER whereas xenografts with low pRb and high p16 were unresponsive breast cancers are clinically distinct subgroups based on their to this therapy. Reduction in pRb levels or RB1 via siRNA knock- prognosis and response to therapy (37, 38). Furthermore, pre- down significantly reduced the sensitivity of TNBC cells to abe- clinical studies show that crosstalk exists between ER and HER2 maciclib. These data support the hypothesis that Rb dependence signaling and activation of either pathway is associated with goes beyond ER-driven breast cancers and that CDK4/6 inhibition resistance to targeting the other pathway (39–41). These data may be effective in other subtypes with the appropriate cyclinD: support the design of the MonarcHER trial (NCT02675231), CDK4/6:Rb activation profile. It is likely that the strong associ- which will evaluate the combination of abemaciclib plus trastu- ation between ER status and abemaciclib response in breast cancer zumab with or without hormonal blockade (fulvestrant) in models and patients (24) is due to the fact that ER is a useful þ þ HER2 /ER breast cancer patients. surrogate marker of downstream dependence on cyclinD:CDK4/ þ The use of CDK4/6 inhibitors beyond ER /HER2 breast 6:Rb:p16 signaling. Selection of patients based on measurement cancer also requires investigation of combinations with stan- of intact Rb pathway signaling as characterized by high tumor dard-of-care cytotoxic chemotherapies. There are preclinical data levels of pRb, low p16, and an absence of cyclin E1 amplification, to suggest that the mechanism of action of molecules that arrest may identify additional populations of patients that could benefit cell cycle may be antagonistic when used with antimitotic agents from abemaciclib therapy and could provide therapeutic benefit (35, 42). To address this question, we first investigated the in vivo in a subset of TNBCs where there are no current, approved targeted activity of abemaciclib in combination with chemotherapy (doc- therapies. etaxel) in HER2-amplfied and TNBC xenografts. In TNBC, cotreat- Of interest, additional factors that associate with response to ment with abemaciclib and docetaxel did not antagonize antitu- abemaciclib in this study include the presence of high levels of mor activity of either single agent. Moreover, the addition of androgen receptor (AR) and the presence of PIK3CA- activating docetaxel to a triple (anti-HER2, anti-ER, and anti-CDK4/6) mutations. It should be noted, however, that these factors track þ þ combination in HER2 /ER xenografts increased the degree of with the luminal breast cancer subtype and as such may not be tumor regressions observed. Analysis of tumor tissues confirmed independent biomarkers of response (19, 45). High levels of that docetaxel did not block the induction of the cell-cycle arrest PTEN protein associated with abemaciclib response indicate that by abemaciclib. In vitro drug-sequencing experiments confirmed intact PI3K pathway signaling may increase the likelihood of that induction of growth arrest by abemaciclib does not block the response to CDK4/6-targeted therapy. inhibition of cell proliferation or the apoptosis induced by PD markers, for the early detection of response to abemaciclib docetaxel or carboplatin. In fact, either simultaneous or sequen- therapy, were also identified in this study using a set of cell-cycle– tial combination of abemaciclib plus either cytotoxic leads to regulated genes. Expression of a core set of five E2F-regulated

www.aacrjournals.org Mol Cancer Ther; 17(5) May 2018 905

O'Brien et al.

genes (RRM2, TOPO2A, MKI67, MCM7, and CDK2), consistently Authors' Contributions tracked with response to abemaciclib in a dose-dependent man- Conception and design: N. O'Brien, D. Conklin, R. Beckmann, C. Mockbee, ner across three independent xenograft studies. Furthermore, the D.J. Slamon dynamic changes in expression of these transcripts tracked with an Development of methodology: D. Conklin, K. Chau, E. von Euw Acquisition of data (provided animals, acquired and managed patients, increase in antitumor response induced by abemaciclib when provided facilities, etc.): T. Luo, K. Chau, J. Thomas, A. Mc Nulty, C. Marchal, combined with fulvestrant, tamoxifen, or trastuzumab. The ulti- E. von Euw mate utility of this marker set should now be evaluated in tissues Analysis and interpretation of data (e.g., statistical analysis, biostatistics, from abemaciclib clinical trials. computational analysis): N. O'Brien, D. Conklin, R. Beckmann, A. Mc Nulty, In this study, we present a comprehensive preclinical profile of S. Hurvitz, C. Mockbee, D.J. Slamon abemaciclib used as a single agent or in combination with Writing, review, and/or revision of the manuscript: N. O'Brien, R. Beckmann, K. Chau, O. Kalous, S. Hurvitz, C. Mockbee, D.J. Slamon standard-of-care therapy in each of the known therapeutic molec- Administrative, technical, or material support (i.e., reporting or organizing ular subtypes of breast cancer. These data further support the data, constructing databases): T. Luo, K. Chau clinical use of abemaciclib as monotherapy as well as a possible þ þ þ combination partner in ER /HER2 , HER2 /ER , and some þ Acknowledgments TNBCs. It is likely that beyond ER /HER2 breast cancer, mea- This study was supported by UCLA funding and a DOD Innovator Award surement of an on-mechanism, validated set of biomarkers to W81XWH-11-1-0104 to Dennis J. Slamon. confirm dependence on Rb -signaling will likely help better select patients that will benefit from abemaciclib-based therapies. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked Disclosure of Potential Conflicts of Interest advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate R. Beckmann has ownership interest (including patents) in Eli Lilly and this fact. Company. S. Hurvitz reports receiving commercial research grant from Lilly. C. Mockbee is an employee of Eli Lilly. D.J. Slamon is a consultant/advisory board member for Eli Lilly. No conflicts of interest were disclosed by the other Received March 29, 2017; revised November 16, 2017; accepted February 16, authors. 2018; published first February 26, 2018.

References 1. Howell A, Osborne CK, Morris C, Wakeling AE. ICI 182,780 (Faslodex): hormone-receptor-positive, HER2-negative metastatic breast cancer that development of a novel, "pure" antiestrogen. Cancer 2000;89:817–25. progressed on previous endocrine therapy (PALOMA-3): final analysis of 2. Osborne CK. Tamoxifen in the treatment of breast cancer. N Engl J Med the multicentre, double-blind, phase 3 randomised controlled trial. Lancet 1998;339:1609–18. Oncol 2016;17:425–39. 3. Osborne CK. Aromatase inhibitors in relation to other forms of endocrine 14.FinnRS,MartinM,RugoHS,JonesS,ImSA,GelmonK,etal.Palbo- therapy for breast cancer. Endocr Relat Cancer 1999;6:271–6. ciclib and letrozole in advanced breast cancer. N Engl J Med 2016;375: 4. Slamon DJ, Leyland-Jones B, Shak S, Fuchs H, Paton V, Bajamonde A, et al. 1925–36. Use of chemotherapy plus a monoclonal antibody against HER2 for 15. Hortobagyi GN, Stemmer SM, Burris HA, Yap YS, Sonke GS, Paluch- metastatic breast cancer that overexpresses HER2. N Engl J Med 2001; Shimon S, et al. Ribociclib as first-line therapy for HR-positive, advanced 344:783–92. breast cancer. N Engl J Med 2016;375:1738–48. 5. Verma S, Miles D, Gianni L, Krop IE, Welslau M, Baselga J, et al. Trastu- 16. Prall OW, Sarcevic B, Musgrove EA, Watts CK, Sutherland RL. Estrogen- zumab emtansine for HER2-positive advanced breast cancer. N Engl J Med induced activation of Cdk4 and Cdk2 during G1-S phase progression is 2012;367:1783–91. accompanied by increased cyclin D1 expression and decreased cyclin- 6. Cuzick J, Sestak I, Baum M, Buzdar A, Howell A, Dowsett M, et al. Effect of dependent kinase inhibitor association with cyclin E-Cdk2. J Biol Chem anastrozole and tamoxifen as adjuvant treatment for early-stage breast 1997;272:10882–94. cancer: 10-year analysis of the ATAC trial. Lancet Oncol 2010;11:1135–41. 17. Harbour JW, Luo RX, Dei Santi A, Postigo AA, Dean DC. Cdk phosphor- 7. Vogel CL, Cobleigh MA, Tripathy D, Gutheil JC, Harris LN, Fehrenbacher L, ylation triggers sequential intramolecular interactions that progressively et al. Efficacy and safety of trastuzumab as a single agent in first-line block Rb functions as cells move through G1. Cell 1999;98:859–69. treatment of HER2-overexpressing metastatic breast cancer. J Clin Oncol 18. Timms JF, White SL, O'Hare MJ, Waterfield MD. Effects of ErbB-2 over- 2002;20:719–26. expression on mitogenic signalling and cell cycle progression in human 8. Berns K, Horlings HM, Hennessy BT, Madiredjo M, Hijmans EM, Beelen K, breast luminal epithelial cells. Oncogene 2002;21:6573–86. et al. A functional genetic approach identifies the PI3K pathway as a major 19. Cancer Genome Atlas N. Comprehensive molecular portraits of human determinant of trastuzumab resistance in breast cancer. Cancer Cell breast tumours. Nature 2012;490:61–70. 2007;12:395–402. 20. Foster SA, Wong DJ, Barrett MT, Galloway DA. Inactivation of p16 in 9. Goel S, Wang Q, Watt AC, Tolaney SM, Dillon DA, Li W, et al. Overcoming human mammary epithelial cells by CpG island methylation. Mol Cell Biol therapeutic resistance in HER2-positive breast cancers with CDK4/6 inhi- 1998;18:1793–801. bitors. Cancer Cell 2016;29:255–69. 21. Gelbert LM, Cai S, Lin X, Sanchez-Martinez C, Del Prado M, Lallena MJ, 10. Miller TW, Balko JM, Fox EM, Ghazoui Z, Dunbier A, Anderson H, et al. et al. Preclinical characterization of the CDK4/6 inhibitor LY2835219: ERalpha-dependent E2F transcription can mediate resistance to estrogen in-vivo cell cycle-dependent/independent anti-tumor activities alone/in deprivation in human breast cancer. Cancer Discov 2011;1:338–51. combination with gemcitabine. Invest New Drugs 2014;32:825–37. 11. Miller TW, Hennessy BT, Gonzalez-Angulo AM, Fox EM, Mills GB, Chen H, 22. Patnaik A, Rosen LS, Tolaney SM, Tolcher AW, Goldman JW, Gandhi L, et al. Hyperactivation of phosphatidylinositol-3 kinase promotes escape et al. Efficacy and safety of abemaciclib, an inhibitor of CDK4 and CDK6, from hormone dependence in estrogen receptor-positive human breast for patients with breast cancer, non-small cell lung cancer, and other solid cancer. J Clin Invest 2010;120:2406–13. tumors. Cancer Discov 2016;6:740–53. 12. O'Brien NA, Browne BC, Chow L, Wang Y, Ginther C, Arboleda J, et al. 23. Finn RS, Crown JP, Ettl J, Schmidt M, Bondarenko IM, Lang I, et al. Efficacy Activated phosphoinositide 3-kinase/AKT signaling confers resistance to and safety of palbociclib in combination with letrozole as first-line treat- trastuzumab but not lapatinib. Mol Cancer Ther 2010;9:1489–502. ment of ER-positive, HER2-negative, advanced breast cancer: expanded 13. Cristofanilli M, Turner NC, Bondarenko I, Ro J, Im SA, Masuda N, et al. analyses of subgroups from the randomized pivotal trial PALOMA-1/ Fulvestrant plus palbociclib versus fulvestrant plus placebo for treatment of TRIO-18. Breast Cancer Res 2016;18:67.

906 Mol Cancer Ther; 17(5) May 2018 Molecular Cancer Therapeutics

Preclinical Activity of Abemaciclib in Breast Cancer

24. Finn RS, Dering J, Conklin D, Kalous O, Cohen DJ, Desai AJ, et al. PD 34. Choi YJ, Li XY, Hydbring P, Sanda T, Stefano J, Christie AL, et al. The 0332991, a selective cyclin D kinase 4/6 inhibitor, preferentially inhibits requirement for cyclin D function in tumor maintenance. Cancer Cell proliferation of luminal estrogen receptor-positive human breast cancer 2012;22:438–51. cell lines in vitro. Breast Cancer Res 2009;11:R77. 35. Roberts PJ, Bisi JE, Strum JC, Combest AJ, Darr DB, Usary JE, et al. Multiple 25. Turner NC, Ro J, Andre F, Loi S, Verma S, Iwata H, et al. Palbociclib in roles of cyclin-dependent kinase 4/6 inhibitors in cancer therapy. J Natl hormone-receptor-positive advanced breast cancer. N Engl J Med 2015; Cancer I 2012;104:476–87. 373:209–19. 36. Witkiewicz AK, Cox D, Knudsen ES. CDK4/6 inhibition provides a potent 26. O'Brien NA, McDonald K, Tong L, von Euw E, Kalous O, Conklin D, et al. adjunct to Her2-targeted therapies in preclinical breast cancer models. Targeting PI3K/mTOR overcomes resistance to HER2-targeted therapy Genes Cancer 2014;5:261–72. independent of feedback activation of AKT. Clin Cancer Res 2014;20: 37. Hurvitz SA, Andre F, Jiang Z, Shao Z, Mano MS, Neciosup SP, et al. 3507–20. Combination of everolimus with trastuzumab plus paclitaxel as ﬁrst-line 27. Konecny GE, Pegram MD, Venkatesan N, Finn R, Yang G, Rahmeh M, et al. treatment for patients with HER2-positive advanced breast cancer Activity of the dual kinase inhibitor lapatinib (GW572016) against HER-2- (BOLERO-1): a phase 3, randomised, double-blind, multicentre trial. overexpressing and trastuzumab-treated breast cancer cells. Cancer Res Lancet Oncol 2015;16:816–29. 2006;66:1630–9. 38. Nahta R, O'Regan RM. Therapeutic implications of estrogen receptor 28. TateSC,CaiS,AjamieRT,BurkeT,BeckmannRP,ChanEM,etal.Semi- signaling in HER2-positive breast cancers. Breast Cancer Res Treat 2012; mechanistic pharmacokinetic/pharmacodynamic modeling of the anti- 135:39–48. tumor activity of LY2835219, a new cyclin-dependent kinase 4/6 39. Gluck S, Arteaga CL, Osborne CK. Optimizing chemotherapy-free survival inhibitor, in mice bearing human tumor xenografts. Clin Cancer Res for the ER/HER2-positive metastatic breast cancer patient. Clin Cancer Res 2014;20:3763–74. 2011;17:5559–61. 29. Infante JR, Cassier PA, Gerecitano JF, Witteveen PO, Chugh R, Ribrag V, 40. Osborne CK, Shou J, Massarweh S, Schiff R. Crosstalk between estrogen et al. A phase I study of the cyclin-dependent kinase 4/6 inhibitor ribociclib receptor and growth factor receptor pathways as a cause for endocrine (LEE011) in patients with advanced solid tumors and lymphomas. Clin therapy resistance in breast cancer. Clin Cancer Res 2005;11(2 Pt 2): Cancer Res 2016;22:5696–705. 865s–70s. 30. DeMichele A, Clark AS, Tan KS, Heitjan DF, Gramlich K, Gallagher M, et al. 41. Xia W, Bacus S, Hegde P, Husain I, Strum J, Liu L, et al. A model of acquired þ CDK 4/6 inhibitor palbociclib (PD0332991) in Rb advanced breast autoresistance to a potent ErbB2 tyrosine kinase inhibitor and a therapeutic cancer: phase II activity, safety, and predictive biomarker assessment. Clin strategy to prevent its onset in breast cancer. Proc Natl Acad Sci U S A 2006; Cancer Res 2015;21:995–1001. 103:7795–800. 31. Dickler MN TS, Rugo HS, Cortes J, DierasV,PattDA,WildiersH,etal. 42. Dean JL, McClendon AK, Knudsen ES. Modiﬁcation of the DNA damage Results from a phase II study of abemaciclib, a CDK4 and CDK6 response by therapeutic CDK4/6 inhibition. J Biol Chem 2012;287: þ inhibitor, as monotherapy, in patients with HR /HER2 breast cancer, 29075–87. after chemotherapy for advanced disease. J Clin Oncol 2016;34(suppl): 43. Konecny GE, Winterhoff B, Kolarova T, Qi J, Manivong K, Dering J, et al. abstract 510. Expression of p16 and retinoblastoma determines response to CDK4/6 32. Sledge GW Jr, Toi M, Neven P, Sohn J, Inoue K, Pivot X, et al. MONARCH 2: inhibition in ovarian cancer. Clin Cancer Res 2011;17:1591–602. þ abemaciclib in combination with fulvestrant in women with HR / 44. Lukas J, Herzinger T, Hansen K, Moroni MC, Resnitzky D, Helin K, et al. HER2 advanced breast cancer who had progressed while receiving endo- Cyclin E-induced S phase without activation of the pRb/E2F pathway. crine therapy. J Clin Oncol 2017;35:2875–84. Genes Dev 1997;11:1479–92. 33. Lenferink AEG, Busse D, Flanagan WM, Yakes FM, Arteaga CL. ErbB2/neu 45. Park S, Koo J, Park HS, Kim JH, Choi SY, Lee JH, et al. Expression of kinase modulates cellular p27(Kip1) and cyclin D1 through multiple androgen receptors in primary breast cancer. Ann Oncol 2010;21: signaling pathways. Cancer Res 2001;61:6583–91. 488–92.

www.aacrjournals.org Mol Cancer Ther; 17(5) May 2018 907

Preclinical Activity of Abemaciclib Alone or in Combination with Antimitotic and Targeted Therapies in Breast Cancer

Neil O'Brien, Dylan Conklin, Richard Beckmann, et al.

Mol Cancer Ther 2018;17:897-907. Published OnlineFirst February 26, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-17-0290

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2018/02/24/1535-7163.MCT-17-0290.DC1

Cited articles This article cites 44 articles, 19 of which you can access for free at: http://mct.aacrjournals.org/content/17/5/897.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/17/5/897.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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/17/5/897. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.