Pivarubicin Is More Effective Than Doxorubicin Against Triple-Negative Breast Cancer In Vivo.

Triple-negative breast cancer (TNBC) is unresponsive to antiestrogen and anti-HER2 therapies, requiring the use of cytotoxic drug combinations of anthracyclines, taxanes, cyclophosphamide, and platinum compounds. Multidrug therapies achieve pathological cure rates of only 2040%, a consequence of drug resistance and cumulative dose limitations necessitated by the reversible cardiotoxic effects of drug therapy. Safer and more effective treatments for TNBC are required to achieve durable therapeutic responses. This study describes the mechanistic analyses of the novel anthracycline, pivarubicin, and its in vivo efficacy against human primary TNBC. Pivarubicin directly activates PKCd, triggers rapid mitochondrial-dependent apoptosis, and circumvents resistance conferred by overexpression of P-glycoprotein, Bcl-2, Bcl-XL, and Bcr-Abl. As a consequence, pivarubicin is more cytotoxic than doxorubicin against MDA-MB-231, and SUM159 TNBC cell lines grown in both monolayer culture and tumorspheres. Comparative in vivo efficacy of pivarubicin and doxorubicin was performed in an orthotopic NSG mouse model implanted with MDA-MB-231 human TNBC cells and treated with the maximum tolerated doses (MTDs) of pivarubicin and doxorubicin. Tumor growth was monitored by digital caliper measurements and determination of endpoint tumor weight and volume. Endpoint cardiotoxicity was assessed histologically by identifying microvacuolization in ventricular cardiomyocytes. Primary tumors treated with multiple rounds of doxorubicin at MTD failed to inhibit tumor growth compared with vehicle-treated tumors. However, administration of a single MTD of pivarubicin produced significant inhibition of tumor growth and tumor regression relative to tumor volume prior to initiation of treatment. Histological analysis of hearts excised from drug- and vehicle-treated mice revealed that pivarubicin produced no evidence of myocardial damage at a therapeutic dose. These results support the development of pivarubicin as a safer and more effective replacement for doxorubicin against TNBC as well as other malignancies for which doxorubicin therapy is indicated.

INTRODUCTION

Triple-negative breast cancer (TNBC) is a highly aggressive subtype that neither expresses estrogen receptors (ERs) and progesterone receptors nor overexpresses epidermal growth factor 2 receptor (HER2). As a consequence, TNBC is unresponsive to anti-estrogen and anti-HER2 therapies and is treated with systemic cytotoxic drug combinations of anthracyclines, such as doxorubicin, taxanes, cyclophosphamide, and platinum compounds1–3. More targeted therapies are undergoing clinical trials4. Despite initial sensitivity to chemotherapy, TNBC patients experience lower overall disease-free intervals compared with patients whose tumors express sex steroid hormone receptors5, with an overall pathological complete response (pCR) of only 20–40%6. The limited cytotoxic efficacy of chemotherapy is likely due to multiple mechanisms of drug resistance, as well as cell senescence and cytoprotective autophagy1. Further, the irreversible cardiotoxic effects of anthracyclines and, to a lesser extent, other chemotherapeutics are well established and limit the cumulative doses of drugs that can be administered to achieve a curative outcome7,8.

Therefore, a critical unmet need exists for safer and more effective treatments for TNBC that eliminate drug-resistant cell subpopulations without producing cardiotoxicities, thereby reducing the probability of recurrent disease and irreversible cardiac damage. N-Benzyladriamycin-14-pivalate (pivarubicin; AD 445) was designed and developed as a chemically stable congener of the experimental antitumor agent N-benzyladriamycin-14-valerate (AD 198)9. AD 198 had been previously shown to competitively bind to the C1b (diacylglycerol binding) regulatory domain of conventional and novel isoforms of protein kinase C (PKC) in the cytoplasmic compartment of mammalian cells10–12. AD 198 is functionally distinct from doxorubicin in its ability to trigger rapid, mitochondrial-dependent apoptosis through PKC-delta (PKCd) activation in a manner that circumvents multiple mechanisms of cellular drug resistance12–16. Further, through the specific activation of PKCe in mammalian cardiomyocytes, AD 198 confers cardioprotection against reperfusion injury following global ischemia and doxorubicin-induced cardiac damage17,18. However, AD 198 is labile to rapid ester hydrolysis of the valerate moiety, resulting in the formation of N-benzyladriamycin (AD 288), a catalytic inhibitor of topoisomerase II with reduced ability to circumvent resistance mediated by multidrug transport proteins or antiapoptotic protein overexpression12,19.

In this study, we will confirm the stability of the tertiary trimethylester moiety of pivarubicin against hydrolysis and determine whether: 1) pivarubicin rapidly triggers apoptosis in a PKCd-dependent manner, 2) these functional characteristics confer therapeutic superiority to pivarubicin over doxorubicin using an orthotopic xenograft model of the aggressive MDA-MB-231-LM2 human TNBC cells implanted in immunodeficient NSG mice, and 3) TNBC tumor growth inhibition by pivarubicin is achieved in the absence of histological evidence of cardiac damage.

MATERIALS AND METHODS

Chemicals and Biologicals

Pivarubicin, originally designed by Dr. Mervyn Israel (University of Tennessee Health Science Center, Memphis, TN, USA), was synthesized by Dr. John Rimoldi (University of Mississippi, Oxford, MS, USA) using a previously described protocol20,21. Doxorubicin HCl, rottlerin, and all antibodies were purchased from Sigma-Aldrich (St. Louis, MO, USA). For in vitro experiments, doxorubicin and pivarubicin were dissolved in dimethyl sulfoxide (DMSO). The final maximum DMSO concentration used for in vitro drug treatments (1% for 72 h) was not cytotoxic. IL-3-dependent 32D.3 murine myeloid cells, CCRF-CEM human lymphoblastic leukemia cells and multidrug-resistant variants (generous gift of Dr. William T. Beck, St. Jude Children’s Research Hospital, Memphis, TN, USA), 293 embryonic human kidney cells and HL-60 human acute myeloid leukemia cells transfected with Bcr-Abl, Bcl-XL, or empty expression vectors (generous gift of Dr. Kapil Bhalla, University of Miami, Coral Gables, FL, USA) were maintained as previously described12,14,22,23. K562 human chronic myelogenous leukemia and LNCaP human prostate cancer cells were purchased from ATCC (Manassas, VA, USA) and maintained as described by the vendor. LNCaP/Bcl-2 cells were the generous gift of Dr. Ralph Buttyan (Columbia University, New York City, NY, USA). Luminescent MDA-MB-231-LM2 (LM-2; metastatic lung subpopulation isolated from human TNBC MDA-MB-231 cells transduced with eIF1a-Luc2-puro lentivirus) were generously provided by Dr. Yibin Kang (Princeton University, Princeton, NJ, USA) and maintained in culture as previously described24. PKCd siRNA25 and scrambled variant were obtained from Qiagen (Hilden, NRW Germany).

Drug Biotransformation Analysis

Quantitative and qualitative determination of AD 198 and pivarubicin biotransformation was determined by reversed-phase HPLC as described previously26,27.

Fluorescence Microscopy

32D.3 cells were grown in suspension culture in the absence of a drug for 24 h prior to analysis. Cells at a density of 1 × 106/ml were exposed to 5 μM doxorubicin or 1 μM pivarubicin for 1 h and then harvested, washed, and resuspended in phosphate-buffered saline, pH-7.2 (PBS). Nuclear counterstaining of pivarubicin-treated cells was performed by treatment of cells with 16 μg/ml of bisbenzimide for 1 h. Drug autofluorescence was observed with an Olympus BH-2 phase-contrast microscope with a mercury UV light source under UV illumination (red: excitation filter, 530–560 nm; barrier filter, 580 nm; blue: excitation filter, 340–390 nm) at a magnification of 1,000×.

Analysis of Apoptosis

Detection of DNA fragmentation in apoptotic cells by the TUNEL assay and immunoblot analyses of cytochrome c release were performed as described previously12.

PKC Inhibition, Cell Viability Determinations, and Immunoblot Analysis

Rottlerin treatment of cells, cell viability analysis by MTT28, and immunoblot identification of protein expression were performed and described previously12.

Injection of Cells Into the Murine Mammary Fat Pad

Monolayer LM2 cells were trypsinized and resuspended in DMEM media containing 10% FBS and 1× antibiotic/antimitotic. Cell concentration was adjusted to yield 2.5 × 105 cells for each 10-μl injection in PBS. After preparation for injection, cells were kept on ice at all times. Cells were surgically implanted into the left and right inguinal mammary glands of 4-week-old female NSG mice (NOD/SCID IL2Rγ−/−; #5557; Jackson Laboratory, Bar Harbor, ME, USA) bred in-house. Mice were anesthetized with 1.2% avertin via intraperitoneal (IP) injection prior to surgery. Mice were also injected subcutaneously between shoulder blades with rimadyl for pain relief on the day of surgery and on the day following surgery. Primary tumor growth prior to and during drug treatment was monitored initially by manual palpation for tumor appearance, then twice weekly using digital calipers. Tumor volume was calculated by the formula: volume = (width2 × length)/2. When tumors attained a volume of approximately 150–190 mm2, cohorts of 10 randomized drug-naive, female NSG mice (8 ± 1 weeks old, 23 ± 2 g) were administered the maximum tolerated dose (MTD) of pivarubicin, doxorubicin, or the equivalent volume of vehicle (70% sterile saline, 15% ethanol, 15% Cremophor EL) as a bolus IP injection in parallel experiments. Tumor volumes were measured in a blinded manner. All mice were maintained on a 7904-irradiated high-fat diet (Teklad, Madison, WI, USA). Food/water was provided ad libitum. Animal health was monitored by visual observation supplemented by body condition scoring as well as weight loss to monitor endpoint (greater than 20%) measured two to three times/week. All animal procedures were approved by the Institutional Animal Care and Use Committee at the University of Tennessee Health Science Center as accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care.

At termination of experiments, mice were sacrificed humanely by CO2 inhalation followed by cervical dislocation.

Imaging Mice Using In Vivo Imaging System

Mice were imaged using Xenogen In Vivo Imaging System® Lumina from PerkinElmer (Waltham, MA, USA). Stock D-luciferin firefly potassium salt (122799; Perkin Elmer) was prepared in Dulbecco’s-PBS (DPBS) at a concentration of 30 mg/ml and sterile filtered with 0.2-μm syringe filter. Luciferin was further diluted to 15 mg/ml in DPBS prior to injection. Mice were injected IP with 200 μl of luciferin. Primary tumors were imaged by placing the mice in dorsal recumbency using luminescent setting of auto exposure, field of view D with a resolution setting of 4 (medium binning), 10 min after injection with luciferin, approximately 5 min following anesthetization with isoflurane. Data are reported as total flux in photons per second (p/s) using the Living Image software.

Cardiotoxicity Assessment

Drug-mediated cardiotoxicity was assessed in both nontumor-bearing female NSG mice administered three doses (IP every 2 weeks) of the MTD of doxorubicin, pivarubicin, or the equivalent volume of solvent only and in hearts excised from tumor-bearing mice treated as described in the Results section. Body weights were monitored every 3 days. Two weeks after the final dose, mice were sacrificed, and the hearts were excised immediately. Intact ventricular myocardia were fixed for a minimum of 24 h in 10% formalin phosphate buffered to pH 7.0 and then carefully sectioned into 2- to 3-mm-thick slices before being dehydrated in graded ethanol and cleared in xylene prior to embedding in paraffin at 58°C. Sections (4-μm thick) were mounted on glass slides, deparaffinized in xylene, and stained in the routine fashion with Mayer’s hematoxylin and eosin. Slide labels were blinded to evaluator. Myocardial lesions were evaluated by routine light microscopy and scored with regard to the severity and extent of damage29:

Degree of severity (S)—0, no evidence of histological changes; 1, sarcoplasmic microvacuolization and/or inclusions (interstitial or cellular edema); 2, as in 1 plus sarcoplasmic macrovacuolization or atrophia, necrosis, fibrosis, endocardial lesions, and thrombi.

Degree of extension (E)—0, no lesions; 0.5, less than 10 single altered myocytes on the whole-heart section; 1, scattered single altered myocytes; 2, scattered small groups of altered myocytes; 3, spread small groups of altered myocytes; 4, confluent groups of altered myocytes; 5, most of cells damaged.

Total cardiotoxicity score/animal = S × E, and mean total score (MTS) for each treatment group was MTS = Σ(S × E)/number of animals.

RESULTS

Pivarubicin was designed to be a hydrolytically stable congener of our previously developed anthracycline antitumor compound, AD 198. In contrast to the straight-chain five-carbon valerate moiety at C-14 appended to the anthraquinone A ring through an esterase-labile ester linkage in AD 198, pivarubicin contains a tertiary trimethylester (pivalate) moiety that likely sterically hinders enzyme-mediated ester hydrolysis, yet retains the three-dimensional configuration that mimics the C1b regulatory domain ligands, diacylglycerol and phorbol 12-myristate 13-acetate11 (Fig. 1). Resistance of the pivalate moiety to hydrolysis is confirmed following treatment of multiple mammalian cell lines with pivarubicin and AD 198, followed by qualitative and quantitative analysis of drug content by reversed-phase HPLC26,27. While AD 198 is subject to approximately 50% biotransformation to N-benzyladriamycin (AD 288) 8 h after drug uptake into cells, less than 10% of pivarubicin is initially biotransformed and shows little time-dependent conversion to AD 288 (Table 1).

Figure 1

Isosteric similarity of pivarubicin and diacylglycerol (DAG). Bold lines identify the putative pharmacophores (C1b binding site) for (A) pivarubicin, (B) diacylglycerol (DAG), and (C) phorbol 12-myristate 13-acetate (PMA).

Table 1

Esterase Resistance of Pivarubicin

	% Parent Compound
	32D.3		293		J774.2
	1 h	8 h	1 h	8 h	1 h	8 h
AD 198*	90%	56%	80%	50%	90%	50%
Pivarubicin*	100%	96%	98%	94%	90%	90%

1 μM drug at 37°C for 1 h/7 h in drug-free medium.

By possessing a hydrolytically stabile C-14 moiety, we determined the functional characteristics of pivarubicin in cell types (CCRF-CEM, 32D.3, LNCaP, HL-60) that were used in the previous mechanistic analyses of AD 19812–14 to determine if hydrolytic stability altered the function of pivarbicin compared with AD 198. As predicted, pivarubicin retains the functional characteristics initially described for AD 198, including rapid cellular uptake and localization almost exclusively in the perinuclear region of the cytoplasm (Fig. 2). Once distributed into microsomal membranes, pivarubicin triggers rapid, mitochondrial-dependent apoptosis that is independent of cell cycle arrest (Fig. 3). Marked release of cytochrome c into the cytosol as an indicator of mitochondrial depolarization and dysfunction is detected within 5 h of drug treatment, while DNA fragmentation, as detected by TUNEL staining, is abundant by 6 h.

Figure 2

Cellular localization of pivarubicin. 32D.3 cells at a density of 1 × 106/ml were exposed to 5 μM doxorubicin or 1 μM pivarubicin for 1 h. Nuclear counterstaining of pivarubicin-treated cells was performed with 16 μg/ml of bisbenzimide for 1 h. Fluorescence microscopy was performed as described in the Materials and Methods section. Images’ original magnification: 1,000×.

Figure 3

Pivarubicin-mediated rapid apoptosis. (A) Cytochrome c release by pivarubicin. 32D.3 cells were exposed to 5 μM pivarubicin for 1 h and then incubated in drug-free medium for an additional 4 h. Cells were then fractionated to isolate the cytosolic fraction, subjected to immunoblotting, and treated with anti-cytochrome c monoclonal antibody (1:250) for 2 h, followed by a 1-h treatment with 1:1,000 dilution of horseradish peroxidase-conjugated goat anti-mouse second antibody as described previously12. Proteins were detected by chemiluminescence. (B) Detection of DNA fragmentation in pivarubicin-treated 32D.3 cells by TUNEL assay. Cells were treated with 5 μM pivarubicin for 1 h, washed twice in warm PBS, incubated in drug-free medium for 6 h, and then prepared for 3′-biotinylation of fragmented DNA using the TUNEL assay procedure. Individual fields of cells were detected for both DNA 3′-end labeling (green) or total cellular DNA staining by propidium iodide (red). Composite image is representative of three independent experiments.

Consistent with the computer modeling and binding studies of N-benzylanthracyclines with varying C-14 acyl chain length and conformation with the C1b regulatory of PKC10,11, pivarubicin-induced cytotoxicity is dependent on PKCd activation (Fig. 4). Within 4 h of treatment with 5 μM pivarubicin, PKCd translocates from the cytosolic to the membrane cellular fraction, consistent with PKCd activation (Fig. 4A). Consistent with this finding, inhibition of PKCd activity also selectively impedes pivarubicin cytotoxicity. Treatment of cells with the IC50 concentration of doxorubicin produces a progressive decrease in the viable cell population down to 50% by 24 h of drug treatment. Cotreatment of cells with rottlerin, a selective inhibitor of the proapoptotic PKCd, does not alter doxorubicin cytotoxicity. However, while pivarubicin produces 50% cell kill within 6 h of treatment, rottlerin markedly delays pivarubicin 50% cell kill to 24 h (Fig. 4B). Likewise, downregulation of PKCd expression in human LNCaP prostate cancer cells by PKCd siRNA transfection inhibits pivarubicin-mediated cell kill, with a 2.5-fold increase in pivarubicin concentration required to achieve 50% cell kill in PKCd–siRNA-transfected cells, thus supporting PKCd activation by pivarubicin as the trigger for apoptosis (Fig. 4C).

Figure 4

Pivarubicin cytotoxicity is mediated through PKC-delta (PKCd) activation. (A) 32D.3 cells, suspended in RPMI-1640 medium/10% FCS/IL-3 at a density of 5 × 105 cells/ml, were treated with either DMSO (C) or 5 μM pivarubicin (PIV) for 4 h prior to harvesting, cell fractionation, and immunoblot analysis of PKCd as described previously14. (B) 32D.3 cells were pretreated with 10 μM rottlerin for 2 h prior to exposure to 5 μM drug for 1 h at 37°C. Cells were pelleted and washed twice in large volumes of warm PBS, then resuspended in fresh, drug-free medium containing 10 μM rottlerin at 5 × 105 cells/ml and incubated at 37°C for up to 72 h. At indicated times, aliquots of cells were withdrawn and stained with trypan blue. Viable cells were scored based on the exclusion of stain and gross morphological appearance. (C) 32D.3 cells were transfected with PKCd siRNA or scrambled siRNA as described in the Materials and Methods section and assessed for pivarubicin cytotoxicity as in (A). Each datum point represents the mean and standard error of at least three independent determinations, each consisting of 300–500 cells per count, when possible.

In addition to its novel mechanisms of cytotoxic action, pivarubicin circumvents multiple mechanisms of cellular drug resistance. CCRF-CEM cells selected for resistance to vinblastine (CEM/VLB-10 and CEM/VLB-100) overexpress P-glycoprotein, exhibit 10.6- and 269-fold resistance to vinblastine22, and exhibit 12- and 73-fold resistance to doxorubicin, respectively. However, VLB-10 cells exhibit no resistance to pivarubicin, and VLB-100 cells are only threefold resistant. Overexpression of transfected antiapoptotic proteins Bcl-2 in LNCaP human prostate cancer cells (Fig. 5A) and Bcl-XL in HL-60 human promyelocytic leukemia cells (Fig. 5B) does not impede the rate of pivarubicin-induced apoptosis. Further, the oncogenic fusion protein, Bcr-Abl, had no inhibitory effect on pivarubicin cytotoxicity in HL-60 cells, in contrast to doxorubicin (Fig. 5B). In total, these cell-based results indicate that, similar to its more chemically labile congener, AD 198, the rapid uptake of pivarubicin in mammalian cells, which is unaffected by P-glycoprotein expression, results in cytoplasmic localization of drug, and rapid translocation and activation of PKCd to trigger mitochondrial-dependent apoptosis in a manner that does not require prior cell cycle arrest and is not impeded by the overexpression of antiapoptotic proteins.

Figure 5

Pivarubicin circumvents multiple mechanisms of MDR. (A) LNCaP cells transfected with Bcl-2 expression vector were assessed for doxorubicin or pivarubicin cytotoxicity, compared to cells transfected with empty vectors. Cells were exposed to 5 μM drug for 1 h, washed, and further incubated for 72 h in drug-free medium. Viability was determined by the MTT assay. (B) Pivarubicin cytotoxicity in HL-60 cells transfected with Bcr-Abl or Bcl-XL expression vectors were assessed for doxorubicin or pivarubicin cytotoxicity and compared as described for (A). Each datum point represents the mean and standard error of at least three independent determinations.

Based upon the ability of pivarubicin to circumvent multiple mechanisms of resistance against doxorubicin, we then examined the efficacy of pivarubicin against TNBC in comparison to doxorubicin. Figure 6A shows the results of in vitro cytotoxicity of pivarubicin and doxorubicin in two basal TNBC cell lines, MDA-MB-231 and SUM15930. Both cell lines exhibit a biphasic response to doxorubicin in MTT assays, with IC50 concentrations for MDA-MB-231 and SUM159 cells of 1 μM and 4 μM, respectively. However, the remaining cell subpopulations were significantly less sensitive to doxorubicin, ultimately shifting the IC90 concentrations to >20 μM in both cell lines. In contrast, the response to pivarubicin treatment was largely monophasic in both cell lines, with IC50 concentrations of 2.5 μM and 1.5 μM, and IC90 concentrations of 4.5 μM and 1.7 μM in MDA-MB-231 and SUM159 cells, respectively. Consistent with monolayer cell culture results, pivarubicin was significantly more effective than doxorubicin in preventing secondary tumorsphere formation by MDA-MB-231 cells, with complete inhibition with 2.5 μM pivarubicin, but only partial inhibition by 2.5 μM doxorubicin (Fig. 6B). Similar results were observed with SUM159 tumorspheres (not shown).

Figure 6

Pivarubicin is more cytotoxic than doxorubicin against triple-negative breast cancer (TNBC) monolayer cells. (A) MDA-MB-231 and SUM-159 cells were maintained under adherent monolayer growth conditions. Upon reaching 50% confluency, cells were treated for 48 h with the dose range of doxorubicin or pivarubicin as indicated. Cell viability was corrected for death observed in vehicle-only-treated cells (<1% ethanol final well concentration). (B) MDA-MB-231 monolayer culture was plated at 6,500 cells per well into a 24-well ultralow adhesion dish in tumorsphere medium with either vehicle, doxorubicin (DOX), or pivarubicin (PIV) and cultured for up to 7 days prior to obtaining digital images. Bar graph represents the mean number of tumorspheres greater than 50 mm in diameter/well (n > 3).

The superiority of pivarubicin over doxorubicin in our in vitro findings led us to investigate whether pivarubicin demonstrates therapeutic superiority in vivo using an orthotopic xenograft model of luminescent MDA-MB-231 cells (LM2) engrafted into the left and right inguinal mammary glands of female NSG mice. Once achieving a primary tumor volume of 150–190 mm3, mice were randomized into three cohorts for treatment with the MTD of pivarubicin or doxorubicin, or the equivalent volume of vehicle per body weight.

The MTD of pivarubicin and doxorubicin, as determined in 8-week-old nontumor-bearing female NSG mice, was defined as the dose given IP (rapid push in 70% sterile saline, 15% EtOH, 15% Cremophor EL) once every 2 weeks (two doses) that produced a maximum reversible 20% decrease in mean body weight. Based on this criterion, the MTD for doxorubicin was 1.4 mg/kg (reversible mean weight loss of 22%). The MTD for pivarubicin was 26.6 mg/kg, producing a reversible mean weight loss of 20%.

Primary LM2 tumors treated with either a single IP dose of 1.4 mg/kg or a subsequent escalated dose of 1.6 mg/kg of doxorubicin 14 days later failed to exhibit reduced growth compared with vehicle treatment, as indicated by in vivo tumor volume measurements during treatment (Fig. 7A) or endpoint volume (Fig. 7B) and weight (Fig. 7C) determinations of excised tumors. Luminescence comparison of doxorubicin- and vehicle-treated primary tumors (Fig. 7D) demonstrated similar flux, suggesting that both groups of tumors had comparable quantities of live cells. Further dosing with doxorubicin was precluded by excessive and irreversible weight loss in tumor-bearing mice.

Figure 7

Doxorubicin treatment of TNBC-bearing NSG mice. (A) LM2 cells were implanted into the 4R and 4L cleared mammary fat pads of female NSG mice and permitted to proliferate for 26 days until a tumor volume range of 125–190 mm3 was reached based on caliper measurements. Doxorubicin was administered IP rapid push in 70% sterile saline, 15% EtOH, 15% Cremophor EL at 14-day intervals, with tumor volumes measured at indicated days postimplantation. Dosing was suspended at 47 days due to tumor size. At endpoint, mice were euthanized and primary tumors resected for volume (B), weight (C), and luminescence (D) measurements. Endpoint results are expressed as the mean ± standard error of 10 mice per group. Statistical significance was determined by the Wilcoxon rank-sum test.

In contrast, a single IP dose of pivarubicin at 26.6 mg/kg produced a mean 40% decrease in tumor volume compared with the mean volume prior to drug administration, indicating not only inhibition of further tumor growth but also regression of tumor mass. This is compared with a 3.5-fold increase in tumor volume in mice treated with vehicle alone (Fig. 8A). As opposed to nontumor-bearing NSG mice, a single MTD of pivarubicin in tumor-bearing NSG mice resulted in a poorly reversible 20% mean weight loss, precluding additional dosing. As a consequence, we performed a second, independent dosing experiment in which tumor-bearing NSG mice were treated with pivarubicin in two doses of 16.5 mg/kg IP at a 2-week interval (Fig. 8B), after which the primary tumors were removed, then measured for volume (Fig. 8C) and weight (Fig. 8D). Endpoint measurements indicated that two rounds of 16.5 mg/kg of pivarubicin produced a 50% decrease in tumor volume and 40% decrease in tumor weight compared with vehicle treatment.

Figure 8

Pivarubicin treatment of TNBC-bearing NSG mice. LM2 cells were implanted into the 4R and 4L cleared mammary fat pads of female NSG mice and permitted to proliferate for 22–24 days until a tumor volume range of 125–190 mm3 was reached based on caliper measurements. Pivarubicin was administered IP rapid push in 70% sterile saline, 15% EtOH, 15% Cremophor EL as either a single dose at 26.6 mg/kg (A) or in two 16.5-mg/kg doses at a 14-day interval (B) with tumor volumes measured at indicated days postimplantation. Dosing was suspended at 47 days due to tumor size in vehicle-treated mice. At endpoint, mice were euthanized and primary tumors resected for volume (C) and weight (D) measurements. Results are expressed as the mean ± standard error of 10 mice per group. Statistical significance was determined by the Wilcoxon rank-sum test.

Taken together, these results indicate that while the MTD of doxorubicin is subtherapeutic against a rapidly growing and aggressive human TNBC orthotopic xenograft, pivarubicin not only inhibited tumor proliferation but also produced tumor regression.

Given the limitation of doxorubicin in the treatment of tumors due to irreversible cardiotoxicity as a consequence of the total cumulative dose of drug, it was next determined whether pivarubicin produced evidence of cardiotoxic damage at the total dose of drug administered to NSG mice that produced tumor growth inhibition. Histological evidence of ventricular damage in hearts of drug- or vehicle-treated NSG mice was assessed using the Bertazzoli test, originally designed to assess doxorubicin-induced cardiomyocyte damage29. Analyses were performed on hearts from two independent experiments: 1) hearts that were excised at dosing endpoint from tumor-bearing NSG mice receiving two IP doses of pivarubicin at 16.5 mg/kg, doxorubicin at 1.4 mg/kg then 1.6 mg/kg, and an equivalent volume of vehicle alone and 2) nontumor-bearing female NSG mice receiving three IP doses of pivarubicin at 16.5 mg/kg, doxorubicin at 1.4 mg/kg then 1.6 mg/kg × 2, and an equivalent volume of vehicle alone at 2-week intervals.

As shown in Table 2, nontumor-bearing mice (n = 9) given a cumulative dose of pivarubicin that was demonstrated to inhibit primary TNBC tumor growth exhibited no evidence of histologic damage to ventricular cardiomyocytes, as measured by both severity of damage to individual cells and extent of the lesion. Similarly, 9/10 hearts from tumor-bearing NSG mice treated with pivarubicin at a cumulative tumor growth inhibitory dose showed no evidence of myocardial damage. In contrast, doxorubicin treatment of nontumor-bearing mice at its subtherapeutic MTD resulted in detectable damage to cardiomyocytes in 4/9 hearts, based on severity. Doxorubicin-treated tumor-bearing mice exhibited less cardiotoxicity (2/9 hearts), but received a lower cumulative dose than nontumor-bearing mice. Paradoxically, vehicle-treated mice from both experiments (tumor bearing, n = 18; nontumor bearing, n = 10) exhibited evidence of myocardial damage in 4/18 and 2/9 hearts, respectively. Figure 9 shows representative stained thin sections of ventricular cardiomyocytes from nontumor-bearing mouse hearts showing visible evidence of microvacuolization. These results indicate that at a cumulative dose that inhibits TNBC tumor growth, pivarubicin produces no detectable cardiac damage.

Table 2

Cardiotoxicity Analysis of Nontumor-Bearing and Triple-Negative Breast Cancer-Bearing Mice

	Nontumor Bearing			Tumor Bearing
	Vehicle	DOX	PIV	Vehicle	DOX	PIV
Severity
0	7/9	5/9	9/9	13/18	7/9	9/10
0.5	0/9	0/9	0/9	0/18	0/9	0/10
1.0	2/9	4/9	0/9	5/18	2/9	1/10
Total score
0	7/9	7/9	9/9	14/18	8/9	9/10
0.5	2/9	2/9	0/9	4/18	1/9	0/10
1.0	0/9	0/9	0/9	0/18	0/9	1/10

Values are number of hearts with indicated score per total number evaluated. DOX, doxorubicin; PIV, pivarubicin.

Figure 9

Representative stained thin sections: cardiotoxicity analysis of nontumor-bearing NSG mice. Images show evidence of microvacuolization (arrows) in both vehicle- and doxorubicin-treated mice.

DISCUSSION

The design of more effective treatments for TNBC continues to be challenging, owing to the 1) absence of exploitable receptor targets, in contrast to ER+/PR+ and HER2+ breast tumors, which limits options for targeted therapy31, and 2) the emergence of refractory cancer cells possessing broad-spectrum resistance, which limits the efficacy of current chemotherapeutic agents2. By exploiting an alternative mechanistic strategy of foregoing the traditional targets of cytotoxic chemotherapy to directly trigger apoptosis via PKCd activation, we report here on the potential superiority of the functionally novel anthracycline, pivarubicin, over the standard-of-care drug, doxorubicin, in the treatment of TNBC. Against primary human MDA-MB-231 TNBC tumors in an orthotopic mouse model, doxorubicin failed to inhibit tumor growth, while pivarubicin not only inhibited growth but also produced tumor regression after a single administration of its MTD. Further, tumor growth inhibition was achieved without the detectable histological damage to ventricular cardiomyocytes often observed at therapeutic doses of doxorubicin in clinical settings, suggesting both improved antitumor efficacy and safety of pivarubicin.

Our current approach to improve chemotherapy with functionally novel anthracycline congeners has been previously successful. Valrubicin (Valstar®) is a cytoplasmic-targeted 14-O-acyl trifluoroacetyl anthracycline, designed and developed in-house32, that is currently FDA approved for intravesicular administration for human superficial bladder carcinoma, as well as in the preclinical stages of development for the treatment of TNBC, psoriasis, and acne33–35. The pivarubicin congener, AD 198, is currently in preclinical development for canine osteosarcoma36, while the mixed function hybrid anthracycline, AD 31237, is undergoing clinical trials against canine B-cell lymphoma. AD 198 has been more extensively evaluated in terms of cytotoxic mechanisms, noncardiotoxicity, cardioprotection, and in vivo antitumor efficacy12,13,17,38. However, the hydrolytic lability of the valerate moiety of AD 198 to yield the functionally distinct congener, AD 288, is both an asset and a liability to antitumor activity. While the biotransformation of AD 198 to AD 288 confers a multifunctional characteristic to AD 198, AD 288 cytotoxicity is inhibited by mechanisms of resistance that do not impede AD 198 or pivarubicin resistance. The sustained stability of pivarubicin ensures a prolonged ability of the drug to remain active against tumor cells that are resistant to chemotherapy through increased expression of multidrug transport proteins or antiapoptotic Bcl-2 proteins, in contrast to AD 19838. Therefore, pivarubicin was selected as our drug of choice for further in vivo evaluation against TNBC.

Pivarubicin possesses unique structural and functional characteristics that promote rapid intracellular accumulation unaffected by P-glycoprotein overexpression and nonnuclear localization in the perinuclear cytoplasm. This unconventional localization of an anthracycline eliminates nuclear DNA damage and subsequent cell cycle arrest as requisite events for cell kill and results in direct activation of cytoplasmic PKCd, leading to mitochondrial-dependent apoptosis that is uninhibited by Bcl-2 overexpression. This latter characteristic is particularly significant in light of the high level of Bcl-2 expression in MDA-MB-231 cells observed by our labs (not shown), as well as by others39–41 and can explain the improved antitumor efficacy of pivarubicin over doxorubicin. Bcl-2 expression impedes doxorubicin-induced apoptosis in TNBC cells42 as well as in a variety of other tumor cell lines43, but has no effect on pivarubicin cytotoxicity. Growth of MDA-MB-231 cells in contact with extracellular matrix proteins has been reported to further enhance Bcl-2 expression, increasing resistance to doxorubicin42. Claudin-low TNBC tumor cells surviving in vivo after neoadjuvant therapy demonstrate the overexpression of both Myc and the antiapoptotic Bcl-2 family protein, Mcl-1, to cooperatively confer drug resistance44 through enhanced mitochondrial oxidative phosphorylation to enhance cancer stem cell viability45. Given the ability of pivarubicin to circumvent the antiapoptotic effects of Bcl-2 family member proteins and the ability of the pivarubicin congener, AD 198, to inhibit Myc expression46,47, it is likely that pivarubicin would be unaffected by the emergence of this mechanism of resistance.

Doxorubicin cytotoxicity is produced, in part, by inhibition of topoisomerase II DNA ligase activity, resulting in the accumulation of double-stranded breaks (DSBs), repairable largely by nonhomologous recombination48. Inactivating BRCA1/2 mutations, occurring in 20% of TNBC tumors, result in reduced DNA double-strand break (DSB) repair and increased tumor sensitivity to anthracyclines4,49 and PARP-inhibitory drugs2. Breast cancer cell lines, such as MDA-MB-231, with normal BRCA but mutant p53 exhibit decreased sensitivity to DNA-damaging agents50. The cytoplasmic localization of pivarubicin and its targeting of C1b domain-containing enzymes, such as PKC and RacGRP, significantly reduce the influence of changes in DSB repair potential on drug cytotoxicity, in contrast to doxorubicin. Additionally, we have previously shown that pivarubicin cellular effects occur independently of ATM or p53 activities51.

Resistance to anthracyclines and taxanes has also been established in TNBC cell lines through the expression of multidrug transport proteins MDR1 and MRP-1, -5, and -652,53, as has been extensively observed in a variety of human tumor cells exposed to these classes of drugs both in vitro and in vivo54,55. While MDA-MB-231 cells express P-gp or MRP only following selective pressure with anthracyclines and other cytotoxic drugs that are transport substrates56, the ability of pivarubicin to circumvent transported-mediated resistance indicates its utility as a salvage therapy for refractory tumors as well as first-line therapy. However, the results presented in this study suggest that LM2 cells, which have not been specifically selected for drug resistance, respond rapidly to pivarubicin but not doxorubicin, pointing to the potential of pivarubicin as first-line treatment.

The precise mechanism of in vivo tumor growth inhibition by pivarubicin is not yet confirmed. While the cytotoxic action of pivarubicin is linked to PKCd-mediated activation of apoptosis, PKCd activity in MDA-MB-231 cells has been linked with the promotion of cell survival57,58 and tumor cell migration through MAPK activation59,60, but is also linked to cell cycle arrest and apoptosis through PKCd-dependent p21 or p27 activity61,62. We have previously reported with AD 198 that rapid apoptosis associated with AD 198-mediated PKCd activation in HL-60 cells occurs simultaneously with the phosphorylation of ERK1/2, STAT5, and Jun14. This suggests that while these anthracyclines have the ability to activate proliferative signaling in a PKC-dependent manner, proapoptotic signaling, nevertheless, predominates. Thus, in LM2 cells, the net effect of pivarubicin treatment is inhibition of tumor growth.

In addition to the limited efficacy of doxorubicin due to multiple mechanisms of multidrug resistance, the increased risk of irreversible cardiac damage by anthracycline therapy is well established and is an impediment to curative anthracycline-based therapy for cancer patients63. Compounding this risk is the established cardiotoxic potential of taxanes and alkylating agents64, which are administered in combination with doxorubicin. As a consequence of the current impediments to curative anthracycline-based therapy, two distinct strategies have been adopted to improve treatment: modification of anthracycline delivery in combination with conventional agents, or identification and exploitation of novel targets in TNBC. Pegylated liposomal doxorubicin (PLD) has been proposed as an alternative for TNBC patients for whom further anthracycline treatment is contraindicated, based on the comparable antitumor efficacy and reduced cardiotoxic potential of PLD31,65. However, while PLD treatment resulted in reduced cardiotoxicity in TNBC patients as well as other, non-life-threatening adverse effects, there were no significant improvements in any therapeutic parameters compared with doxorubicin. As a consequence, PLD has been considered as only noninferior to doxorubicin in clinical trials31.

As an alternative to the conventional mechanisms of action of DNA damage and mitotic spindle inhibition, a wide variety of new targets in TNBC are currently under evaluation in phases I–II clinical trials, including inhibitors of DNA repair enzymes (PARP), inhibitors of proliferative signaling components (mTOR, AKT, MEK), 3) immunomodulatory targets (PD-1), and hormone receptors (AR)2,31. While drugs that target these sites have resulted in pCR of 39% with bevacizumab (anti-VEGF)and up to 85% with pembrolizumab (PD-1 blocker) in early clinical trials, all targeted agents have been administered in combination with conventional agents, including doxorubicin to achieve therapeutic success2.

While the curative potential of pivarubicin against TNBC is not yet firmly established, this proof-of-principle study suggests that TNBC tumor eradication with pivarubicin can be achieved with optimal dosing and scheduling. A bolus MTD of pivarubicin was clearly effective in reducing TNBC tumor volume, but with >20% mean body weight loss that was poorly recoverable, but without accompanying morbidities, precluded further dosing to achieve additional tumor regression and eradication, as per animal care guidelines. This, combined with the highly aggressive and invasive nature of LM2 cells, with rapid and early spread to axillary lymph nodes, liver, and lungs66 suggests that fractional dosing on a daily basis would be more pharmacologically and therapeutically advantageous, as we have previously observed with AD 19838 and valrubicin (M. Israel, personal communication). We noted an apparent difference in MTD for pivarubicin in tumor-bearing versus nontumor-bearing mice, whereas the doxorubicin MTD remained unchanged. This discrepancy points to the need for a more comprehensive understanding of pivarubicin pharmacokinetics in tumor- versus nontumor-bearing animals as well as further investigation into optimal dosing regimens for pivarubicin, both of which are planned.

The absence of histological evidence of cardiotoxicity at therapeutic doses of pivarubicin, including microvacuolization (Table 2, Fig. 9) and myofibrillar loss (not shown), would not necessarily be predictable, given the multiple possible mechanisms of anthracycline-induced cardiotoxicity in addition to the generation of ROS from the anthraquinone ring structure67–69. Indeed, preliminary morphological analysis revealed ventricular wall thinning and cavity enlargement in some pivarubicin-treated hearts from tumor-bearing mice described in Figure 8 and enlargement of hearts from nontumor-bearing mice treated with pivarubicin and to a lesser extent with doxorubicin (not shown). However, based on the close functional similarity of pivarubicin to AD 198 and our previous observations that AD 198 is both noncardiotoxic and functionally cardioprotective through activation of PKCe in rodent cardiomyocytes despite generation of ROS comparable to doxorubicin17,18, pivarubicin possesses the potential to be noncardiotoxic. It is worth noting in our results that vehicle treatment, alone, produced microvacuolization similar to doxorubicin/vehicle and greater than pivarubicin/vehicle treatment. While Cremophor EL/ethanol has been previously reported to produce a decrease in mitochondrial respiration70, there have been no reports of Cremophor EL/ethanol-induced microvacuolization in cardiomyocytes in either our laboratories or elsewhere. Given the marginal histological evidence of cardiotoxicity with doxorobucin in our studies, the differences in cardiotoxicity between vehicle and doxorubicin may not be significant. The nearly complete absence of microvacuolization in pivarubicin/vehicle-treated hearts could be a consequence of predicted cardioprotective action of pivarubicin, but this remains to be established, as does the structural and functional cardiac effects of pivarubicin through a more in-depth analysis.

In summary, the initial success in demonstrating the potential superiority of pivarubicin over doxorubicin against human TNBC in a mouse xenograft model is the first step in the clinical development of pivarubicin as a safer and more effective replacement for doxorubicin in TNBC combination chemotherapy. This preclinical proof-of-principle will be followed up by pivarubicin dosing optimization analysis of pivarubicin efficacy against TNBC metastases in our human TNBC xenograft model as a prelude to eventual clinical development.