Consecutive Day HSP90 Inhibitor Administration Improves Efficacy in Murine Models of KIT-Driven Malignancies and Canine Mast Cell Tumors

Heat shock proteins (HSP) are molecular chaperones integral for regulating multiple aspects of posttranslational protein homeostasis for numerous client proteins, including protein kinases and transcription factors involved in cell proliferation, signaling, and oncogenesis (1, 2). Heat shock protein 90 (HSP90) is important for maintenance and stabilization of known oncoproteins, including KIT, MET, and EGFR, among others. Upregulation of HSP90 expression and function has been well described in cancer cells (3–5), and its activity is generally considered crucial for tumor cell survival, supporting its targeting for therapeutic intervention.

Considerable progress in the development of HSP90 inhibitors (HSP90i) has occurred over the past 15 years. Early geldanamycin-based HSP90i displayed modest activity and were associated with significant clinical toxicity, as well as potential multidrug efflux and formulation problems, limiting their clinical development (6–9). STA-1474 (Synta Pharmaceuticals Corp) is a novel water-soluble resorcinol-containing compound metabolized to STA-9090, a potent HSP90i that binds to the N-terminus of the HSP90 ATP-binding domain, resulting in destabilization and degradation of HSP90 client proteins. HSP90i have shown activity against several types of cancers both in vitro and in vivo, including models of canine mast cell tumor (MCT) and canine osteosarcoma (10, 11). STA-9090 (ganetespib) demonstrated activity in a canine MCT mouse xenograft model and induced growth inhibition, apoptosis, and downregulation of pKIT, KIT, and AKT in both KIT-dependent and KIT-independent tumors (10).

More recently, in a phase I clinical trial, STA-1474 was administered to 25 dogs with spontaneous tumors exploring 3 different dosing schemes (12). Measurable objective responses were observed in dogs with MCT (n = 3), osteosarcoma, melanoma, and thyroid carcinoma (n = 1 each), for an objective response rate of 24% (6/25). Data from this trial suggested that longer drug infusion times (8 hours vs. 1 hour) were potentially associated with greater biologic activity, evidenced by a higher objective response rate and sustained plasma levels of STA-9090. When administered over 8 hours, STA-9090 plasma levels were maintained for approximately 12 hours, compared with only 5 to 6 hours when administered as a 1-hour infusion. Low-objective response rates have been reported in phase I and II human clinical trials of STA-9090 in which drug was administered as a 1-hour infusion once or twice per week (72 hours apart), although several patients experienced stable disease (13–15). Taken together, these data suggest that longer plasma exposures of STA-9090 may be associated with greater biologic activity. It is well established that inhibition of HSP90 induces the heat shock response, whereby HSF-1 is released from HSP90 and induces the synthesis of HSP90 and other heat shock proteins. Thus, one potential mechanism for the improved efficacy with longer infusion times may be a result of inhibition of this newly synthesized material.

Given the clinical challenges associated with prolonged intravenous infusions, this study sought to determine whether 2 consecutive days of treatment with STA-1474 was superior to other dose regimens with respect to target modulation and clinical response to therapy by providing a strategy to inhibit both baseline HSP90 as well as HSP90i-induced HSP90. As such, the overriding objective of this study was to expand upon the initial phase I study findings and identify the optimal dosing schedule of STA-1474 in dogs with spontaneous MCT, using both response to therapy and duration of KIT downregulation as biomarkers of efficacy.

The H1975 and Kasumi cell lines were obtained from the American Type Culture Collection and maintained according to standard techniques and used within 2 years of receipt. Primary antibodies were obtained from Cell Signaling Technology, with the exception of GAPDH (Santa Cruz Biotechnology) and HSP70 (Enzo life sciences). Ganetespib and STA-12-7888 (D3-ganetespib) were synthesized by Synta Pharmaceuticals Corp.

Following in vitro assays, tumor cells were disrupted in lysis buffer (Cell Signaling Technology) on ice for 10 minutes. For the pharmacodynamic analysis, xenograft tumors (average volume of 100–200 mm³) were excised, cut in half, and flash frozen in liquid nitrogen. Each tumor fragment was lysed in 0.5 mL of lysis buffer using a FastPrep-24 homogenizer and Lysing Matrix A (MP Biomedicals). Lysates were clarified by centrifugation and equal amounts of protein resolved by SDS-PAGE before transfer to nitrocellulose membranes (Invitrogen). Membranes were blocked with Starting Block T20 Blocking Buffer (Thermo Fisher Scientific) and immunoblotted with the indicated antibodies. Antibody–antigen complexes were visualized using an Odyssey system (LI-COR).

H1975 cells, cultured in RPMI-1640 and 10% FBS, were seeded at a density of 3 × 10⁵ cells per well in 6-well plates. Twenty-four hours later, cells were treated with ganetespib as indicated and incubated at 37°C. Cells were washed twice in cold PBS then lysed in cold HSP90-binding buffer (20 mmol/L HEPES pH 7.3, 1 mmol/L EDTA, 100 mmol/L KCl, 5 mmol/L MgCl, 0.01% v/v NP-40, 0.5 mg/mL bovine gamma globulin, 1 mmol/L TCEP) by incubation on ice for 10 minutes followed by three freeze/thaw cycles. Lysates were clarified by centrifugation at 14,000 × g. To remove unbound ganetespib, lysates were passed over 40 K MWCO size exclusion columns (Thermo Fisher Scientific). To titrate unoccupied HSP90-binding sites, 10 μmol/L of a deuterated form of ganetespib (D3-ganetespib) was added to eluates and incubated at 4°C for 2 hours then passed over a size exclusion column to remove unbound D3-ganetespib. Total protein from flow through was quantified by BCA protein assay and all samples diluted to 1 mg/mL. The concentrations of ganetespib and D3-ganetespib were measured by LC/MS-MS. A Phenomenex Kinetex column (C18, 30 × 2.1 mm, 2.6 μm) was used with a run time of 3.5 minutes per sample. The following equation was used to calculate the percentage of ganetespib bound to HSP90 (HSP90 occupancy): (ganetespib)/[(ganetespib) + (D3-ganetespib)] × 100.

Female CB.17 (SCID) mice (Charles River Laboratories) at 7 to 12 weeks of age were maintained in a pathogen-free environment and all in vivo procedures were approved by the Synta Pharmaceuticals Corp. Institutional Animal Care and Use Committee. Human GIST882 cells were provided by Dr. Jonathan Fletcher (Dana-Farber Cancer Institute) and implanted subcutaneously at 10 × 10⁶ into mice. Mice bearing established tumors (∼110 mm³) were randomized into treatment groups of 8 and dosed intravenously (i.v.) with vehicle or ganetespib, formulated in DRD (10% DMSO, 18% Cremophor RH 40, 3.6% dextrose), using the schedules indicated. Human H1975 non–small cell lung cancer (NSCLC) cells were purchased from the ATCC, selected to stably express a HIF1α-LUC reporter and implanted at 10 × 10⁶ into mice. Mice bearing established tumors (∼143 mm³) were randomized into treatment groups of 4 and dosed intravenously with vehicle or ganetespib, formulated in DRD, using the schedules indicated. Tumor volumes (V) were calculated by caliper measurements of the width (W), length (L) and thickness (T) of each tumor using the formula: V = 0.5236 (LWT). Tumor growth inhibition was determined as described previously (16). Statistical analyses were conducted using two-way ANOVA followed by Bonferroni posttests.

A phase II randomized open label study of STA-1474 was completed in client-owned dogs with spontaneous MCT greater than 2 cm in size. Dogs were randomized to receive a single dose of STA-1474 intravenously at 6 mg/kg over 1 hour (Cohort A), 6 mg/kg over 8 hours (cohort B), 3 mg/kg over 1 hour on day 1 and day 2 (D1/D2) (cohort C), or 3 mg/kg on day 1 and day 4 (D1/D4; cohort D). This treatment schedule was repeated for a total of 4 cycles. Tumor biopsies, obtained under sedation and local anesthesia using a 3 to 5 mm punch biopsy or 14-gauge Tru-cut needle, were taken at baseline prior to the first treatment with STA-1474 and then 24 and 72 hours after the last treatment of the first cycle. All biopsy samples were flash frozen in liquid nitrogen and stored at −80°C until analysis. Routine hematologic and biochemical profiles were performed once weekly throughout the study period. Dogs were evaluated for clinical toxicities and adverse events at each study visit. Clinical toxicities related to disease progression or other unrelated comorbid conditions were not considered adverse events (AE). All AEs were graded in accordance with established VCOG-CTCAE criteria (17).

The Clinical Research and Advising Committee at the College of Veterinary Medicine and the Institutional Animal Care and Use Committee at The Ohio State University approved this study. Informed consent was obtained from all owners prior to study entry. To be considered for enrollment, dogs were required to have a confirmed diagnosis of MCT. Additional required eligibility criteria included the following: at least 1 year of age at the time of enrollment; adequate organ function as indicated by standard laboratory evaluation; no prior chemotherapy administration or radiotherapy; and no other serious systemic disorder incompatible with the study. Prior use of prednisone was permitted.

STA-1474 was supplied by Synta Pharmaceuticals in 30 mg/mL vials containing either 30 mg or 150 mg of drug. Unopened vials were stored for no longer than 30 days at 4°C and protected from light. STA-1474 was diluted in Plasmalyte-148 in a 250 mL infusion bag for administration and was used within 24 hours of preparation. Each vial was used within 96 hours of seal puncture, and the remaining drug was discarded.

Response assessments were performed at enrollment and once weekly thereafter. Responses were characterized according to the VCOG response evaluation criteria for solid tumors in dogs (v1.0) via serial examination (18). An objective response consisted of either a complete or partial response. A complete response (CR) was defined as complete resolution of all target lesions. Partial response (PR) was defined as ≥30% reduction in the sum of the longest dimensions of the target lesions, taking as a reference the baseline sum of the longest dimensions. Progressive disease (PD) was defined as a >20% increase in the sum of the longest dimensions of the target lesions, taking as a reference the smallest sum of the longest dimensions since treatment initiation, or the appearance of at least 1 new nontarget lesion. Stable disease (SD) was defined as neither sufficient decrease nor increase in target lesions to be considered an objective response or disease progression, respectively. All target lesions, including metastatic lesions, were evaluated using caliper measurements or ultrasonographic evaluation of distant metastasis.

The primary AEs associated with STA-1474 administration identified in the phase I trial were gastrointestinal in nature (12). Therefore, dogs were administered maropitant, ondansetron, and metronidazole prophylactically before and after each STA-1474 infusion. Concomitant medications used to prevent and/or treat clinical toxicities were used at the discretion of the attending clinician and included the following: antiemetics (maropitant, ondansetron, metoclopramide, mirtazapine), gastroprotectants (omeprazole, famotidine, sucralfate, pantoprazole), intravenous fluids, antidiarrheals (metronidazole, loperamide, bismuth subsalicylate, probiotics), antihistamines (diphenhydramine), analgesics (tramadol, butorphanol), hepatoprotectants (S-adenosylmethionine, silybin), antibiotics (amoxicillin clavulanate, cephalexin, doxycycline, cefovecin). Prednisone use was permitted during the study if dogs were already receiving the drug and had demonstrated PD despite treatment, and as a single administration for relief of swelling and erythema associated with MCT biopsy.

MCT biopsies were collected and flash frozen before treatment (0 hour), and at 24 and 72 hours after completion of the STA-1474 i.v. infusion. Frozen biopsies were pulverized into a powder while in liquid nitrogen. Tissue powder was thawed briefly on ice, then resuspended in fresh lysis buffer consisting of 20 mmol/L Tris-HCl pH 8.0, 137 mmol/L NaCl, 10% glycerol, 1% IPEGAL CA-630, 10 mmol/L ethylenediaminetetraacetic acid (EDTA), 1 mg/mL aprotinin, 1 mg/mL leupeptin, 1 mg/mL pepstatin A, 1 mmol/L phenylmethylsulphonyl fluoride, 1 mmol/L sodium orthovanadate, and 10 mmol/L sodium fluoride (all from Sigma-Aldrich). Samples were vortexed, then rocked for 1 hour at 4°C. The protein lysates were collected and quantified using the Bradford assay. Protein (1 mg/mL per sample) was precleared for 1 hour with protein A agarose beads (Roche Diagnostic Corp.), followed by immunoprecipitation with 5 μg of anti-KIT (CD117, Pierce/Thermo Fisher Scientific). Immunoprecipitated protein was separated by SDS-PAGE and transferred to a PVDF membrane. Membranes were incubated overnight with anti-KIT antibody (CD117, Pierce/Thermo Fisher Scientific). The membranes were incubated for 1 hour with goat anti-rabbit horseradish peroxidase linked secondary antibody, washed, and exposed to substrate (SuperSignal West Dura Extended Duration Substrate, Pierce).

In vitro modeling of ganetespib exposure was used to provide insight into the mechanism of action associated with improved biological responses after prolonged HSP90i exposure in vivo. EGFR-mutant (L858R and T790M) H1975 NSCLC cells were exposed to increasing doses of ganetespib for either 1 hour or 24 hours (continuous), followed by a washout and incubation with media and DMSO vehicle for 23 hours. Nearly complete client protein turnover (EGFR, HER2, MET), effector inactivation (pAKT, pSTAT3, pERK), heat shock response (HSP70), and signals of apoptosis (BIM) were observed with constant ganetespib treatment at 50 nmol/L (Fig. 1A). Approximately 30-fold higher ganetespib (1,500 nm) concentration was required to obtain comparable effects after a 1-hour treatment. Similar results were observed in the acute myeloid leukemia line Kasumi-1, which harbors an activating mutation (N822K) in KIT (Fig. 1B), and in H2228 NSCLC cells, H3122 NSCLC cells, SKBR3 breast cancer cells, A375 melanoma cells, and MKN45 gastric cancer cells (data not shown).

We next assessed the kinetics of protein degradation following robust HSP90 inhibition and observed variability among proteins, consistent with normal variability in rates of protein turnover. EGFR (L858R T790M) and MET showed ∼50% turnover after 8 hours of exposure, and nearly 100% turnover by 24 hours in H1975 NSCLC cells (Fig. 1C), concordant with their published half-lives (19–21). Rapid HER2 turnover was observed, with the majority of turnover observed within 4 hours. Consistent with its published half-life, HER2 turnover was nearly complete by 8 hours (22). KIT expression in Kasumi-1 cells appeared to follow this trend, with significant degradation of total and phosphorylated KIT at 4 hours (Fig. 1D), reflecting the published half-lives for the two isoforms (23). Taken together, these results indicate that durable inhibitory effects on the chaperone require either continuous low levels of drug or pulse-dosed high drug levels.

Therapeutic activity of HSP90 inhibition is likely directly related to the time taken for client proteins to return to pretreatment expression levels after their degradation. As such, the pharmacodynamics of receptor tyrosine kinase expression in both H1975 and Kasumi-1 cell line xenografts were investigated. EGFR, HER2, and MET suppression was observed in H1795 xenografts 24 hours after ganetespib exposure and persisted for approximately 5 days, (Fig. 2A). HSP70 levels increased following ganetespib exposure and remained elevated after 5 days. Phosphorylated KIT (Y719) was rapidly degraded (6 hours) in gastrointestinal stromal (GIST) xenograft tumors expressing an activating KIT mutation (K642E) as was the cyclin-dependent kinase CDC2/CDK1 (Fig. 2B). In contrast to EGFR, HER2, MET, and CDC2/CDK1, expression of KIT and levels of pKIT (Y719) returned to near baseline levels within 24 hours after ganetespib administration. These data suggest that in diseases driven by proteins such as EGFR and HER2, weekly treatment with an HSP90i may yield clinical activity whereas in tumors driven by proteins such as KIT, efficacy could be minimal and more frequent dosing or longer exposure times would likely be required for optimal biologic activity.

Why did KIT levels stabilize so quickly after HSP90 inhibition? We hypothesized that the observed rapid stabilization of KIT after HSP90i treatment likely had to be HSP90-dependent. Moreover, given the feedback upregulation of HSP70 following ganetespib exposure (Figs. 1 and 2) perhaps HSP90 upregulation was compensating for the loss of steady state HSP90 activity. We assessed the kinetics of HSP90 expression following ganetespib exposure in vitro. Shown in Fig. 2C, HSP90 was induced in H1975 cells after treatment with a single dose of ganetespib, reaching a maximum at 8 hours and remaining elevated at 24 hours. Based on these results, we expected the percentage of total HSP90 bound by ganetespib following a single pulse dose to decline over time because it would be diluted by the newly synthesized, unbound HSP90. To evaluate this, we first determined the amount of ganetespib required to saturate available HSP90-binding sites, by treating H1975 cells for 1 hour with increasing doses of ganetespib, followed by the addition of a deuterated form of ganetespib (10 μmol/L) to the cell lysate to occupy any unbound sites, an approach similar to that previously reported (24). Approximately 64 to 128 nmol/L ganetespib was sufficient to occupy ∼80% of available binding sites (Fig. 2D) within 5 to 10 minutes of drug exposure (Fig. 2E). These values are approximate as we did not correct for binding to the plate, cells, or proteins in/out of the cell. Next, we investigated whether the total number of HSP90-binding sites would increase with time following ganetespib exposure, based on total HSP90i occupancy. At 6 hours after ganetespib treatment, there was a negligible rise in occupancy, but by 16 hours, the number of binding sites increased 2.3-fold (Fig. 2F). This occupancy could be blocked with the protein synthesis inhibitor cycloheximide. Taken together, these results suggest that although an initial exposure to ganetespib can effectively saturate HSP90-binding sites, upregulation of new HSP90 protein provides a new pool of chaperone; combined with drug clearance (25), this results in a sustained increase in HSP90 expression.

We then asked whether 2 ganetespib treatments, administered 24 hours apart, would trigger another round of HSP90 synthesis. Although a slight rise in HSP70 expression was observed between the 24-hour single exposure and 8 hours after 2 consecutive-day treatments (Fig. 2C), levels of HSP90 remained stable and were not greatly elevated by the second ganetespib treatment, nor was there a substantial rise in new HPS90-binding sites (Fig. 2H). These data support the notion that the capacity to make new HSP90 following HSP90i treatment may be limited with a subsequent, near-term exposure to HSP90i and may provide a strategy for greater HSP90 occupancy, more durable client protein degradation, and ultimately, enhanced efficacy.

To test whether a second, near-term administration of ganetespib would limit the capacity to make new HSP90, ganetespib levels were measured in H1975 cell lysates (after size exclusion column to remove free drug) 4 hours after a single 1-hour treatment or after 2 consecutive days of treatment. Indeed, ganetespib levels were 1.6-fold higher in the latter treatment group (Fig. 2G). HSP90 occupancy by ganetespib in H1975 cells (Fig. 2H) following a single pulse exposure decreased 44% between 8 and 24 hours in relation to the total level of HSP90i (ganetespib plus deuterated ganetespib), corresponding with an increase in new HSP90 synthesis (Fig. 2C). Consistent with new HSP90 synthesis, ganetespib occupied an even greater total number of HSP90-binding sites after 2 consecutive days of dosing; however, these levels remained stable for 48 hours. These results indicate that consecutive ganetespib administration does not promote a secondary induction of HSP90 synthesis and represents a viable mechanism to increase total HSP90 binding/inhibition.

Mice bearing human GIST882 tumor xenografts were treated with vehicle or ganetespib administered as a single dose (100 mg/kg), q12h (50 mg/kg), or on 2 consecutive days (50 mg/kg). These doses have been well documented as safe in mice (below the highest nonseverely toxic dose) and exposures are consistent with what is clinically obtainable and meaningful in humans (14, 25). Tumors initially regressed in all drug treatment groups; however, progression was evident early in the once weekly dosing arm. More frequent ganetespib exposure resulted in biologic activity (SD, PR) at the end of the 2-week study; however, the difference was not statistically significant (Fig. 3A). Similar results were observed in H1975 xenografts (Fig. 3B), where 2 cycles of once weekly ganetespib treatment resulted in SD and 2 cycles of 2 consecutive-day ganetespib treatment promoted significant and durable tumor regression (PR; P = 0.001). These findings are consistent with previously published data demonstrating SD with weekly ganetespib treatment in H1975 tumors and regression when dosed 5 times per week (25).

We had previously evaluated the water-soluble prodrug of ganetespib, STA-1474, in dogs with spontaneous cancers and found that biologic activity was greatest in dogs that received an 8-hour drug infusion. This correlated to sustained (8+ hours) plasma ganetespib levels ranging from 200 to 600 ng/mL, and we hypothesized that this was sufficient to promote additional inhibition of any newly synthesized HSP90 following initial drug inhibition. However, a similar protocol could not be used in humans, given the solvent required for ganetespib, thus limiting infusion times in humans. Therefore, we initiated a clinical trial in dogs with spontaneous MCT to determine whether D1/D2 treatment with an HSP90i (STA-1474) was superior to other dose regimens with respect to objective responses and target modulation.

Twenty-four dogs (n = 6 per treatment group) with MCT were enrolled into this clinical trial. All MCTs were chemotherapy and radiotherapy naïve. Patient demographics are detailed in Table 1. The median age was 8 years, and the median weight was 34.1 kg. Eleven dogs received a short course of corticosteroids (typically one injection or 2–3 days of therapy) during the study period due to swelling associated with MCT biopsies. In mice, short-term corticosteroid use does not affect KIT expression in normal mast cells (26). In dogs with MCT treated with prednisone alone at 40 mg/m² daily, total KIT expression levels did not change based on IHC performed at baseline and day 21 of therapy (27). These data suggest that the short-term use of corticosteroids in this study was unlikely to have affected the assessment of changes in KIT expression following treatment.

Treatment groups were designed to expand upon the phase I study results to determine whether 2 consecutive days of STA-1474 treatment were superior to one 8-hour infusion, with respect to target modulation and objective response to therapy. Two cohorts of dogs (cohorts A and B) were treated at 6 mg/kg STA-1474 over 1 hour or 8 hours (n = 6 each) once per week. An additional 2 cohorts of dogs (n = 6 each; cohorts C and D) were treated at 3 mg/kg STA-1474 over 1 hour on days 1 and 2 (D1/D2), or days 1 and 4 (D1/D4). Each dog was scheduled to receive 4 cycles of STA-1474. Three dogs were withdrawn from the study prior to completion (n = 1 comorbid condition; n = 3 PD).

AEs associated with STA-1474 infusion were mild and primarily gastrointestinal in nature, consisting of diarrhea, vomiting, anorexia, and lethargy (Table 2; Supplementary Table S1). There were no dose-limiting toxicities, defined as any grade 3 or 4 adverse event, related to STA-1474 administration. Liver transaminase elevations occurred in all dose groups. These elevations were considered related to steroid administration or comorbid conditions and were not considered related to STA-1474 administration. Low-grade BUN elevations and anemia occurred in cohorts A, B, and D. These clinical toxicities are well established in dogs with MCT disease and were deemed unlikely to be related to STA-1474. One dog with preexisting IRIS stage II chronic kidney disease (CKD) developed grade 3 elevations in BUN and creatinine during the study period. This dog was hospitalized on fluid therapy for 3 days and was discharged from the hospital upon resolution of the azotemia. This dog was subsequently withdrawn from the clinical trial due to progressive CKD.

Other treatment emergent AEs included regurgitation and hind limb weakness in one dog (cohort C). This dog regurgitated a small amount of water 1 day after the first infusion, and metoclopramide was administered prophylactically for subsequent STA-1474 infusions. At week 3, the same dog was reportedly weak in the hind legs when ambulating at home. No detectable neurologic dysfunction or weakness was detectable while in the hospital or during STA-1474 treatments. The weakness was determined to be most likely related to obesity; however, an adverse event secondary to STA-1474 could not be excluded.

Measurable clinical benefit was observed in all cohorts; however, no durable (≥5 week) objective responses were seen in dogs that received STA-1474 at 6 mg/kg once weekly over 1 hour (Fig. 4A). Three dogs were withdrawn prior to study completion due to PD (n = 2 cohort D; n = 1 cohort A), and 1 dog (cohort B) was removed from the study due to progression of CKD. Dogs that received 3 mg/kg STA-1474 on 2 consecutive days demonstrated biologic activity (100%) with 50% of those dogs experiencing an objective response, when compared with all other dosing groups (Fig. 4A and C).

Immunoprecipitation and Western blotting were performed on flash frozen tumor samples from 3 dogs each in cohort taken predose and 24 and 72 hours after STA-1474 dosing (Fig. 4B). Although some degree of KIT modulation was observed in all cohorts, only those samples from cohort C demonstrated durable loss of KIT that persisted at the 72-hour time point. These data are consistent with the clinical responses and confirming the notion that durable HSP90 inhibition following treatment with STA-1474 requires more sustained drug exposures.

A variety of HSP90 inhibitors have demonstrated activity and promise in vitro; however, translating this into clinical benefit in human cancer patients has proven difficult, highlighting the challenges associated with inhibition of this pathway in a therapeutically efficacious manner. Several mechanisms have been proposed for the observed limited biologic activity in patients, including concurrent upregulation of cochaperones that stabilize client proteins, low tumor reliance on the client protein, and suboptimal client protein inhibition due to conventional dosing regimens (6). This study used a relevant large animal model of a KIT-driven tumor, recapitulating the molecular heterogeneity and environmental context of HSP90 inhibition in human cancers, to validate an improved dosing schedule, identified through preclinical work in human and murine cell lines with known KIT dysregulation, associated with objective responses and modulation of a driver oncoprotein (KIT) following STA-1474 treatment. The use of preclinical models with distinct tumor histology but with a similar molecular aberration strengthens the preclinical evidence used to determine the optimal dosing regimen of HSP90i treatment.

In cancer cell lines, the dose and regimen of HSP90i treatment has a significant impact on the degree and duration of client protein degradation. Consistent with drug concentrations necessary to saturate chaperone binding sites, constant exposure to ganetespib (50–125 nmol/L) promoted degradation of HSP90 clients and inactivation of downstream effectors. In contrast, transient ganetespib treatment required 1–2 log more drug for comparable effects. This may be due in part to the need to occupy and inhibit nascent HSP90 induced by the heat shock response, requiring nonspecifically retained excess drug to act as a drug reservoir.

Variations in client turnover and stabilization were observed in murine xenograft models following ganetespib treatment. For example, rapid turnover of pKIT was observed following single-dose ganetespib administration in mice bearing GIST xenograft tumors. However, expression levels of the kinase returned to near baseline levels 24 hours after ganetespib administration, resulting in poor tumor control in vivo in animals receiving once weekly dosing. Similar pKIT kinetics were observed in patient tumor biopsies during a phase II trial of ganetespib administered at the maximum tolerated dosed (200 mg/m²) once weekly for 3 out of 4 weeks (28). Although disease stabilization was observed in 12/23 patients, durable inhibition of pKIT and its downstream pathways was not observed, supporting our preclinical finding that transient suppression of KIT activation occurs with once weekly HSP90i administration.

In contrast, mutant EGFR was fully degraded 24 hours after ganetespib administration in the H1975 NSCLC tumors and remained at low levels for the duration of the 5-day study. In this model, weekly ganetespib treatment led to SD. However, CRs were observed in transgenic tumor models, dosing daily with a selective third-generation EGFR inhibitor, suggesting that constant inhibition is required for objective responses (29). Indeed, administering low-dose ganetespib 5 days per week resulted in significant tumor regression in H1975 xenograft models (25). Interestingly, lung tumor xenografts driven by an EML4-ALK fusion were highly responsive to once weekly ganetespib treatment, with objective responses observed at the MTD (150 mg/kg) and SD observed at lower doses (50 mg/kg) (30).

We sought to identify an HSP90i administration schedule that would combine a clinically feasible dosing regimen with the biologic activity of constant HSP90i exposure to inhibit both baseline and drug-induced HSP90, thus extending the duration of client protein degradation. It is firmly established that inhibition of HSP90 triggers a heat shock response, resulting in the induction of heat shock proteins, including HSP90 and HSP70. Our occupancy studies demonstrated that ganetespib could rapidly occupy the majority of cellular HSP90 in short-term studies, supporting previously published research regarding the impact HSP90i have on oncoprotein stability and cell signaling pathways. Importantly, the rise in new HSP90-binding sites after HSP90i exposure was determined to be due to new protein synthesis, and subsequent HSP90i administration occupied both baseline and induced HSP90-binding sites. These effects result in increased ganetespib exposure, and therefore efficacy, as evidenced by the responses observed in the NSCLC and GIST xenograft studies.

The importance of HSP90 inhibitor dose schedule is highlighted in recent clinical studies. For example, a phase II trial of the HSP90 inhibitor 17-AAG (tanespimycin) was associated with minimal HSP90 inhibition and few objective responses in metastatic melanoma (9). Pharmacodynamic results from this study indicated that the limited success was attributable to suboptimal dose schedule, evidenced by limited inhibition of target client proteins and variable pharmacokinetics (9). More recently, a phase I study of STA-1474 in spontaneous canine cancers suggested that prolonged infusions were associated with greater biologic activity (12). In the present study, two 1-hour infusions given 24 hours apart was associated with the greatest objective response rate (50%), supporting the notion that sustained HSP90 inhibition is associated with improved biologic activity. This treatment regimen was also associated with a persistent decrease in KIT in tumor samples 72 hours posttreatment, while less frequent STA-1474 administration was associated with fewer dogs experiencing clinical benefit and persistent KIT expression in tumor samples after treatment. These findings warrant future prospective studies in larger cohorts of patients to validate the results presented herein.

Taken together, our data underscore the importance of determining the dosing regimens that best modulate intratumoral HSP90 function to provide sustained inhibition of cancer cell growth and survival. AEs reported herein were consistent with those reported in the phase I study of STA-1474 in dogs with cancer (12). AEs were primarily gastrointestinal in nature (lethargy, anorexia, vomiting, diarrhea) and resolved with supportive care. EGFR is an HSP90 client protein with a role in intestinal epithelial integrity, and EGFR inhibition has been postulated to be a mechanism of diarrhea associated with HSP90 inhibitors (31–33). Gastrointestinal AEs are the known primary toxicity of STA-1474, and in the present study, prophylactic management of nausea and diarrhea were utilized to help reduce the incidence and severity of gastrointestinal toxicities. Importantly, all dosing regimens were well tolerated in this study, and enhanced toxicity was not apparent due to increased frequency of STA-1474 administration. This is particularly important with respect to geldanamycin derivatives, in which the administration schedule significantly affects drug tolerability (34–36).

In addition, ganetespib has demonstrated synergistic activity with the MET kinase inhibitor crizotinib, both in the setting of crizotinib-sensitive and -resistant MET-driven tumor models (37). These findings raise the question of whether a combinatorial approach to treatment of KIT-driven malignancies could produce synergistic activity and abrogate the development of drug resistance. Although the objective responses observed in this study after STA-1474 treatment demonstrate the potential therapeutic utility of HSP90 inhibition in canine MCT, no dogs experienced a CR to therapy, and further investigation of combinatorial approaches may improve objective responses.

In summary, more frequent exposure to STA-1474 shows evidence of improved biologic activity without an enhanced toxicity profile in a relevant, spontaneous, large animal model of cancer. These findings indicate that more frequent exposure to HSP90 inhibitors will translate into improved biologic activity in people.

D.A. Proia was Director at Synta Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C.A. London, D.A. Proia

Development of methodology: D.A. Proia

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.A. London, J. Acquaviva, D.L. Smith, M. Sequeira, L.S. Ogawa, L.F. Bernabe, M.D. Bear, S.A. Bechtel

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.A. London, J. Acquaviva, D.L. Smith, D.A. Proia

Writing, review, and/or revision of the manuscript: C.A. London, J. Acquaviva, H.L. Gardner, S.A. Bechtel, D.A. Proia

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.A. London, H.L. Gardner, L.F. Bernabe, M.D. Bear, S.A. Bechtel

Study supervision: C.A. London, D.A. Proia

The Veterinary Clinical Research Services Shared Resource at the Ohio State University College of Veterinary Medicine coordinated all aspects of this study including generation of case report forms, collection of samples, collation of data, quality assurance, coordination of other clinical sites, and final quality control on all data from all study sites. This project was supported by the following grants: UL1TR001070 from the National Center for Advancing Translational Sciences and P30CA016058 from the National Cancer Institute to The Ohio State University. Synta Pharmaceuticals provided funding for this clinical trial. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Advancing Translational Sciences, National Cancer Institute or the National Institutes of Health, Synta Pharmaceuticals, or Aldeyra Therapeutics. This clinical trial was directly supported by funds from Synta Pharmaceuticals. This project was supported by the following grants: UL1TR001070 from the National Center for Advancing Translational Sciences and P30CA016058 from the National Cancer Institute to The Ohio State University.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.