A Phase I Dose-Escalation Study of Danusertib (PHA-739358) Administered as a 24-Hour Infusion with and without Granulocyte Colony-Stimulating Factor in a 14-Day Cycle in Patients with Advanced Solid Tumors

Danusertib is a small-molecule 3-aminopyrazole derivative that potently inhibits all Aurora kinase family members (A, B, and C). Aurora kinases are essential for mitosis (1) and cytokinesis with upregulation in many malignancies, including pancreatic (2), colorectal (3), ovarian (3), and esophageal cancer (4). Aurora A is required for spindle assembly and localizes to centrosomes. Aurora B is a chromosome passenger protein required for histone H3 phosphorylation (pH3), chromosome segregation, and cytokinesis (5–7). Several Aurora kinase inhibitors with diverse pharmacologic properties are in clinical trials (8–15).

Danusertib is active in vitro against a wide range of cancer cell lines with submicromolar IC₅₀ values for inhibition of proliferation. Danusertib shows dominant Aurora B kinase inhibition (16) with additional activity against bcr-abl, including the T315I mutation. In vitro studies have shown that danusertib causes a failure of cell division, resulting in polyploidy and reduction in viability. Danusertib inhibits Aurora B phosphorylation at serine 10 of histone H-3, a protein implicated in chromosome condensation. Thus, inhibition of pH3 may be a potential biomarker of danusertib biological activity. Danusertib also has significant antitumor activity in transgenic tumor models with a favorable preclinical safety profile (16); principal target organs of danusertib are the hemolymphopoietic system, gastrointestinal tract, male reproductive organs and kidneys. Renal effects, however, are only seen at high drug exposure. In addition, an increase in peripheral blood pressure was observed after i.v. bolus and 6-hour infusion in rats, but not in dogs.

Based on this broad spectrum of preclinical activity and favorable toxicology profile, we performed a phase I study to evaluate the safety, tolerability, pharmacokinetics, and pharmacodynamics of danusertib in patients with refractory solid tumors. To the best of our knowledge, danusertib was the first Aurora kinase inhibitor in the clinic.

Patients with histologic or cytologic evidence of advanced refractory cancer lacking options for established curative or life-prolonging therapy were eligible. Eligibility criteria also included Eastern Cooperative Oncology Group performance status of ≤1; prior radiation completed ≥2 wk before enrollment and ≤.25% of bone marrow irradiated; life expectancy of ≥12 wk; normal blood pressure (≤140/90 mmHg) with or without treatment; and baseline laboratory data indicating acceptable bone marrow, liver, and kidney function. Patients with previous high-dose chemotherapy requiring hematopoietic stem cell rescue, known brain or leptomeningeal metastasis, active inflammatory bowel disease, partial or complete bowel obstruction or chronic diarrhea, abnormal left ventricular function, severe cardiovascular disease, cardiac dysrhythmias of ≥grade 2, and active infections were excluded. All patients were adults (age >18 y) and gave written informed consent according to local Institutional Review Board and Federal guidelines.

This was a phase I, open-label, nonrandomized, dose-escalation trial conducted at 2 U.S. sites. Danusertib was manufactured by Nerviano Medical Sciences S.r.l. and supplied as 1% (w/v) sterile solution in 20/26 mL type I glass vials. Each vial contained 15 mL of solution (10 mg/mL) corresponding to 150 mg of danusertib. The drug product was stored at 2°C to 8°C, protected from light, and brought to room temperature shortly before use. Danusertib was administered via central line as a 24-h i.v. infusion on day 1 of a 14-d treatment cycle. The starting dose was 45 mg/m², targeted to administer 1/10th of the exposure at maximum tolerated dose (MTD) in the dog, the most sensitive species in toxicology studies. Routine antiemetic prophylaxis was not given. Incidence and severity of adverse events (AE) were coded according to National Cancer Institute Common Terminology Criteria (version 3.0). Response to therapy was monitored by Response Evaluation Criteria in Solid Tumors. Dose-limiting toxicity (DLT) was defined as grade 4 neutropenia lasting >7 d, febrile neutropenia, neutropenic infection, grade 3 thrombocytopenia for >7 d or associated with bleeding, grade 4 thrombocytopenia, or any grade 3/4 nonhematologic toxicity including dose delay by ≥1 wk occurring during cycle 1 attributable to danusertib. The MTD was the highest dose at which less than or equal to one of six patients had a DLT during cycle 1. Patients could continue on danusertib (14-d cycles) in the absence of disease progression or unacceptable toxicity.

A two-stage accelerated titration design was used with initial rapid dose escalation (100% dose increments) until occurrence of DLT during cycle 1, or two patients at any dose level (DL)/any cycle developing grade ≥2 drug-related toxicity. Thereafter, a modified Fibonacci scheme was to be used with dose increments of 50%, 40%, and 33%. Cohorts of three patients (expanded to six if one DLT was observed among the initial three patients) were to be treated with progressively higher doses. The MTD was defined as the highest dose at which no more than 1 of 6 new patients developed DLT during the first course. After the determination of MTD without granulocyte colony-stimulating factor (G-CSF; part 1), we attempted further dose escalation in ≤33% dose increments using G-CSF for primary prophylaxis of hematologic toxicity (part 2). Treatment with G-CSF (5 μg/kg/daily) was started 24 h after the end of danusertib infusion and continued for 10 d (from day 3-12) ending 48 h before the next infusion.

Complete histories, physical examinations, complete blood counts, serum electrolytes, chemistries, and urinalysis were done at baseline and before each cycle of treatment. Laboratory studies were repeated weekly. Coagulation tests (PT, PTT, fibrinogen concentration, fibrin degradation products, and D-dimer) were done at predose, 1 h after the end of infusion and 24 h after the end of infusion in cycle 1 with repetition in subsequent cycles for any clinically significant changes. Twelve-lead electrocardiogram and baseline MUGA scan were done within 28 d before start of dosing with subsequent MUGA scans at the end of cycle 1, at the end of every even cycle up to cycle 6, and then every fourth cycle [if left ventricular ejection fraction (LVEF) decrease was <10%]. Radiological exams (computed tomography, magnetic resonance imaging) were done at baseline and after every three 2-wk cycles of therapy. Criteria for patient discontinuation included progressive disease at any time, unacceptable toxicity, changes in medical status, patient refusal or noncompliance, or lost to follow-up. Patients could receive focal palliative radiotherapy and remain on-study provided radiation-related toxicities (other than xerostomia) had normalized to preradiation levels 2 wk after completion of radiotherapy. If such radiotherapy involved >25% of bone marrow, however, this was considered disease progression.

Plasma samples for evaluation of pharmacokinetics (PK) of danusertib and its main metabolite (see chemical structures in Fig. 1A) were collected during cycle 1 (on days 1, 2, 3, 4, and 8) before, during, and after the end of infusion for a total of 17 samples. For subsequent cycles, PK sampling was done less frequently (three samples: predose, 5 min before and 1 h after the end of each infusion). In patients experiencing hypertensive episodes, additional PK sampling was to be done 1 to 5 min before stopping the infusion. Urine samples were collected at predose and for 72 h after the start of the infusion in cycle 1. Analytes were quantitated using a validated liquid chromatography-tandem mass spectrometry technique (see Supplementary Data A: Bioanalytical Method). Individual PK parameters of clearance, volume of distribution of the terminal phase, half-life, and area under the time-concentration curve (AUC) were determined by noncompartmental analysis using WinNonlin software (version 3.1, Pharsight, Inc.). C_max values were derived from the plasma concentration-time profiles. The elimination rate constant, λ, was estimated by linear regression. AUC was calculated using the linear/logarithmic trapezoidal rule.

Normal skin biopsies (3-4 mm punch) were taken from the upper arm or other hairless area for evaluation of pH3 at baseline and before the end of infusion. All samples were fixed immediately in 10% buffered formalin solution (see Supplementary Data B: IHC Method).

All patients who received at least one dose were considered assessable for safety and efficacy. Safety data were summarized using appropriate descriptive statistics.

Patient characteristics are summarized in Table 1. A total of 57 patients were enrolled and 56 treated between July 2004 and February 2008. Colorectal cancer was the most frequent cancer type (12 and 7 patients in the danusertib and danusertib + G-CSF groups, respectively). All patients had received prior systemic therapy, except for one patient with chondrosarcoma. More than one third had also received radiotherapy.

As shown in Table 2, the initial dose escalation over seven DLs (45, 90, 180, 360, 500, 580, 650 mg/m²) was without G-CSF. Dose escalation through the 500 mg/m² DL was well-tolerated. At the 650 mg/m² DL (30% increment), two patients developed DLTs. Based on PK analysis (30% coefficient of variability in C_max and AUC), we evaluated an intermediate DL of 580 mg/m²; there were two DLTs at this DL. Thus, 500 mg/m² is the danusertib MTD without G-CSF. Based on DLTs of neutropenia, we then amended the protocol to continue dose escalation with G-CSF primary prophylaxis. Sixteen patients were treated at three DLs with G-CSF (580, 750, and 1,000 mg/m²). We did not identify a per protocol MTD with G-CSF but halted dose escalation and declared 750 mg/m² the MTD based on two patients at the 1,000 mg/m² DL with elevated creatinine (azotemia) requiring a several day hospitalization for management with i.v. fluids. Forty patients treated with danusertib alone received a total of 159 cycles. A total of 66 cycles were administered to the 16 patients treated with danusertib with G-CSF (Table 2). The median number of cycles per patient was three with either treatment regimen (range, 1-20 cycles for the danusertib only group and 1-12 cycles for the danusertib with G-CSF group). The median treatment duration was ∼6 weeks in both groups [danusertib, 6.0 weeks (range 1.0-48.9); danusertib with G-CSF, 6.1 weeks (range 0.9-24.0)].

The side effects of danusertib and the number of patients experiencing various grades of treatment-related toxicity (all cycles) are summarized by dose level and maximum Common Terminology Criteria (CTC) grade in Table 3. Toxicities were primarily hematologic with febrile neutropenia as the cycle 1 dose-limiting toxicity in the group treated without G-CSF. Treatment emergent grade 3 to 4 hematologic toxicities occurred at all dose levels tested, with greater frequency at doses ≥360 mg/m². In cycle 1, median time to neutropenic nadir was 8 days with both regimens. Median time to recovery was shorter when danusertib was administered with G-CSF (3 versus 7 days without G-CSF). The most frequent nonhematologic AEs in both groups were fatigue, anorexia, nausea, vomiting, and diarrhea and pyrexia. Most of these AEs were reported at doses ≥360 mg/m² (Table 3A). Additional drug-related grade 3 to 4 events were zoster with neutropenia (one case), rigors with neutropenia (one case), mucositis (2 cases), and increased liver function tests and hypokalemia (1 case each).

Transient hypertensive episodes (defined as SBP > 150, or DBP > 100, or increase by 20 mmHg from preinfusion values) were observed in 20% of patients receiving danusertib only (one case each at 90, 180, and 500 mg/m²; three cases at 360 mg/m², and 2 cases at 580 mg/m²) and in 19% of those receiving danusertib + G-CSF (one case at each DL). Hypertensive episodes were observed mostly during cycle 1, within 12 hours after the start of the infusion, were mild to moderate in severity, not dose-limiting, and did not recur with repeated treatment. Hypertension was reported as a clinical AE (not related to study drug) in one patient treated at 580 mg/m² with G-CSF. Six of the 12 patients experiencing hypertension were already on antihypertensive medication. We cannot exclude a role for danusertib in these hypertensive episodes as the drug does inhibit vascular endothelial growth factor receptor in a submicromolar range in biochemical assays.

LVEF decreases to values below the lower limit of normal were detected by MUGA scan in two cases treated with danusertib alone at the 45 (−8%) and 360 mg/m² DLs (−9%). LVEF reductions were asymptomatic and most recovered during or after treatment without specific intervention. Decreases of ≥15% were not observed.

DLTs are summarized in Table 2. Based on the observed DLTs, the 500 mg/m² DL is the MTD for the 24-h infusion without G-CSF. In the danusertib with G-CSF group, a first cycle DLT (febrile neutropenia) was reported in only one of seven patients treated at 1,000 mg/m² DL.

Drug-related serious AEs (SAE) occurred in 12 patients (21%): 9 (22%) treated with danusertib without G-CSF (in 2, 3, and 4 cases at 500, 580, or 650 mg/m², respectively) and in 3 patients (19%) receiving 1,000 mg/m² with G-CSF. In patients treated without G-CSF, SAEs were mainly hematologic (febrile neutropenia, neutropenia, neutropenic infection, leukopenia, and anemia). Nonhematologic drug-related SAEs included abnormal liver function tests, vomiting, rigors, and pyrexia (one case each). In patients receiving danusertib with G-CSF, drug-related nonhematologic SAEs included anorexia, nausea, renal failure, azotemia, and pyrexia (one case each).

Reasons for study discontinuation were disease progression (73% and 88% without and with G-CSF, respectively) or AEs (17% and 13% without and with G-CSF, respectively). One patient was never treated. The remaining patients completed treatment per protocol Dose reductions included two patients with grade 1 and 2 creatinine increases requiring medical treatment (i.v. hydration) at 1,000 mg/m² with G-CSF.

Danusertib pharmacokinetics parameters are summarized in Table 4. Mean ± SD plasma levels of the compound and its major metabolite at the MTD doses with and without G-CSF are presented in Fig. 1B and C, respectively. The major metabolite observed in PK studies is the N-oxide of the N-methyl piperazine moiety; this has <1% of the activity of the parent compound in the proliferation assays and potency ∼10 times lower than that of danusertib in an A2780 xenograft mouse model. After reaching maximal plasma concentration, danusertib showed a polyexponential decline. The PK of danusertib are characterized by high volume of distribution (average, 13 L/kg) and low-to-moderate plasma clearance (range, 0.33-0.46 L/h/kg). Renal clearance accounted for a small proportion of plasma clearance (∼10%). No differences in half-life were seen across doses (average, 18-26 hours). Plasma levels of the metabolite declined in parallel with those of the parent compound. The metabolite to parent AUC ratio was similar across doses and approximately equal to 1. Plasma concentrations of danusertib and its metabolite observed at selected time points in subsequent cycles were similar to those observed in cycle 1. Systemic exposure to danusertib increased with dose (Fig. 1D). Administration of G-CSF did not influence the PK of danusertib and its metabolite (Table 4).

Pretreatment and on-treatment skin biopsies were analyzed for changes in pH3 by Western Blot and immunohistochemistry (Supplementary Data B; Fig. 2). By Western Blot, most samples were not evaluable due to protein degradation; however, among evaluable samples, a decrease in pH3 was observed in biopsies starting at 500 mg/m². This finding was confirmed by immunohistochemistry evaluations. Although the number of pH3-positive cells detected was low, a trend toward down-modulation of pH3 was seen starting at 500 mg/m².

Objective tumor response was observed in one patient with refractory small cell lung cancer receiving 1,000 mg/m² danusertib + G-CSF (subsequently reduced to 750 mg/m² for grade 2 creatinine elevation). Disease stabilization was observed in 18 of 42 evaluable patients [43%; danusertib, 11 patients (39%); danusertib + G-CSF, 7 patients (50%)]. Prolonged disease stabilization (23.9-52.3 weeks) was observed in four patients (two colorectal cancer, one ovarian cancer, one breast cancer) treated without G-CSF at 180 mg/m², 500 mg/m² (two patients), and 650 mg/m². In the patient with ovarian cancer, tumor masses decreased 27% (Response Evaluation Criteria in Solid Tumors) with a ≥30% decrease in CA125.

We report the results of a phase I trial of danusertib (PHA-739358), a novel pan-aurora kinase inhibitor, in patients with advanced refractory solid tumors. Given their key roles in regulation of mitotic processes during cell division (5, 17, 18) and overexpression in malignancy, Aurora kinases have emerged as a new target for anticancer therapy. Several small molecule aurora kinase inhibitors are under development and have entered clinical development (11, 15, 19–27). Danusertib has inhibitory activity against Aurora A and B, with potent preclinical antitumor activity, a favorable nonclinical safety profile as well as potentially useful effects on other tyrosine kinases such as bcr-abl (16, 28, 29).

The dosing schedule tested was selected based on preclinical toxicity studies in rats and dogs. In these species, a weekly schedule of danusertib showed acceptable toxicity with the major target organs being bone marrow and gastrointestinal tract. When danusertib was given at very high doses and/or with infusion times shorter than 1 hour, cardiovascular and renal toxicity were seen in addition to bone marrow and gastrointestinal toxicity. Sudden death also occurred in animals given bolus infusions. The 24-hour infusion duration was selected for the present study to reduce toxicities thought to be related to C_max. Danusertib doses of ∼420 to 750 mg/m² were predicted in humans to be required to achieve a daily AUC of ∼50 μmol/L·h, the systemic exposure suggested by preclinical pharmacokinetic/pharmacodynamic modeling to be necessary for tumor regression.

The safety profile of danusertib in this clinical study was completely consistent with that expected based on preclinical testing. Danusertib toxicities were primarily hematologic with febrile neutropenia as the DLT, and 500 mg/m² as the MTD without G-CSF. With G-CSF support the recommended phase 2 dose (RP2D) for danusertib when administered as 24-hour i.v. infusion is 750 mg/m². Doses of >1,000 mg/m² DL were not evaluated, because of the emergence in two patients treated at 1,000 mg/m² of clinically significant grade 1 and 2 creatinine increases, requiring hospitalization for i.v. fluids and restoration of baseline kidney function. In a rat model, tubular nephropathy was observed at the 600 and 750 mg/m² i.v. bolus equivalent doses. In repeated dose studies with 6-hour i.v. infusions, increases in urea and creatinine were observed in rats (but not in dogs) at the MTD of 720 mg/m² and were reversible. A satisfactory mechanistic explanation of these effects has not yet been found.

In the range of doses investigated (45-1,000 mg/m²), the systemic exposure to danusertib increased with dose. Danusertib was characterized by an extensive volume of distribution and low-moderate plasma clearance. Renal clearance accounted for ∼10% of total clearance. PK evaluations showed that coadministration of G-CSF did not influence the PK of danusertib and its metabolite. There was limited interpatient and intrapatient variability in PK. Analysis of pH3 in posttreatment skin biopsies showed target modulation at doses ≥500 mg/m².

In conclusion, danusertib can be safely administered to patients with advanced refractory solid tumors. Danusertib 500 mg/m² given over 24 hours in 14-day cycles without G-CSF is the RP2D for solid tumors. Danusertib 750 mg/m² is the RP2D with G-CSF. A prolonged objective response in small cell lung carcinoma and multiple instances of prolonged disease stabilization provide ample justification for further studies of danusertib alone and in combination with other agents. Phase II and III single agent studies without G-CSF are under way in seven types of solid tumors. In addition, clinical studies with G-CSF are being considered.

R. Cohen has received commercial research grants from Pfizer and Nerviano and is a member of a Nerviano Medical Sciences advisory board. The other authors disclosed no potential conflicts of interest.

We thank the patients who participated in this study and their referring doctors, the study coordinators and data managers for their the skilled contributions, as well as P. Carpinelli (NMS, Western Blot assays), C. Pellizzoni (Accelera, PK reports), A. Petroccione (NMS, Biostatistics), D. Zanchetta (NMS, Data Management), and Nancy Malinowski (editorial assistance).