Phase II study of MLN8237 (Alisertib) in advanced/metastatic sarcoma

Advanced sarcoma is an aggressive disease with poor prognosis and limited treatment options. This clinical trial tested the activity of the aurora kinase inhibitor alisertib in patients with advanced sarcoma. The study was based on promising preclinical data which demonstrated the activity of aurora kinase inhibitors in sarcoma models. Alisertib was tested in five separate histologically defined cohorts. Although the study did not meet its primary end point of response rate, progression-free survival was favorable and responses were seen in two patients with metastatic angiosarcoma. Tumor biopsies demonstrated inhibition of both aurora kinase A and B.

Abstract

Background

Aurora kinase A (AURKA) is commonly overexpressed in sarcoma. The inhibition of AURKA by shRNA or by a specific AURKA inhibitor blocks in vitro proliferation of multiple sarcoma subtypes. MLN8237 (alisertib) is a novel oral adenosine triphosphate-competitive AURKA inhibitor.

Patients and methods

This Cancer Therapy Evaluation Program-sponsored phase II study of alisertib was conducted through the Alliance for Clinical Trials in Oncology (A091102). Patients were enrolled into histology-defined cohorts: (i) liposarcoma, (ii) leiomyosarcoma, (iii) undifferentiated sarcoma, (iv) malignant peripheral nerve sheath tumor, or (v) other. Treatment was alisertib 50 mg PO b.i.d. d1–d7 every 21 days. The primary end point was response rate; progression-free survival (PFS) was secondary. One response in the first 9 patients expanded enrollment in a cohort to 24 using a Simon two-stage design.

Results

Seventy-two patients were enrolled at 24 sites [12 LPS, 10 LMS, 11 US, 10 malignant peripheral nerve sheath tumor (MPNST), 29 Other]. The median age was 55 years; 54% were male; 58%/38%/4% were ECOG PS 0/1/2. One PR expanded enrollment to the second stage in the other sarcoma cohort. The histology-specific cohorts ceased at the first stage. There were two confirmed PRs in the other cohort (both angiosarcoma) and one unconfirmed PR in dedifferentiated chondrosarcoma. Twelve-week PFS was 73% (LPS), 44% (LMS), 36% (US), 60% (MPNST), and 38% (Other). Grade 3–4 adverse events: oral mucositis (12%), anemia (14%), platelet count decreased (14%), leukopenia (22%), and neutropenia (42%).

Conclusions

Alisertib was well tolerated. Occasional responses, yet prolonged stable disease, were observed. Although failing to meet the primary RR end point, PFS was promising.

Trial registration ID

NCT01653028.

introduction

Soft tissue sarcomas (STS) and bone sarcomas are heterogeneous malignant tumors of mesenchymal origin. Surgery, often with adjuvant radiation therapy for larger tumors, is the mainstay of treatment; however, many patients develop recurrent or metastatic disease. The standard of care for many years has been doxorubicin or a doxorubicin combination. Response rates, however, are low [1]. Another chemotherapy option for metastatic disease is gemcitabine with docetaxel [2]. Pazopanib, a tyrosine kinase inhibitor, was recently approved and although it delays progression in some patients, responses remain rare [3]. Thus, there is a dearth of active agents and an unmet medical need.

The aurora kinase family of serine/threonine kinases consists of three members, designated aurora kinases A, B, and C [4]. Aurora kinases A and B are expressed in many different cell types, whereas the expression of aurora kinase C is restricted to testicular tissue. Aurora kinase A (AURKA) is located at the centrosome and required for centrosome maturation and division as well as the formation of the mitotic spindle. AURKA inhibition leads to mitotic delays, defects in chromosome segregation, and cell death [5]. AURKA is commonly overexpressed in STS and inhibition of AURKA by shRNA or by a specific AURKA inhibitor inhibits proliferation of STS cells [6].

Alisertib (MLN8237) is a novel oral, adenosine triphosphate-competitive, selective small-molecule inhibitor of AURKA. Alisertib is a reversible AURKA inhibitor with an inhibition constant (Ki) of 0.43 nM. Alisertib inhibited aurora kinase A activity in HCT-116 human colon cancer cells with a half-maximal inhibitory concentration (IC₅₀) of 6.7 nM and was ∼200-fold more selective for AURKA than aurora kinase B (IC₅₀ = 1534 nM) [7]. In addition, alisertib was 250-fold more selective for AURKA when compared with other kinases tested in vitro [8]. In preclinical models, alisertib has shown activity in multiple STS subtypes, including dedifferentiated liposarcoma, leiomyosarcoma, osteosarcoma, Ewing sarcoma, and malignant peripheral nerve sheath tumor (MPNST) [9].

A phase I trial examining the safety and efficacy of alisertib has been completed [10]. Grade 3 and 4 neutropenia with stomatitis was dose-limiting and the recommended phase II dose was 50 mg b.i.d. for 7 days, every 21 days.

methods

The primary end point of this study was to determine the response rate to treatment with alisertib in patients with sarcoma. Secondary end points were progression-free survival (PFS) and overall survival (OS). This multicenter trial was sponsored by the Cancer Therapy Evaluation Program (CTEP) and conducted through the Alliance for Clinical Trials in Oncology (Alliance A091102). Patients were enrolled into one of five histology-defined cohorts and a Simon two-stage design was used for each cohort (Figure 1).

Figure 1.

Progression-free survival and overall survival by cohort.

Key eligibility criteria included age ≥18 years, measurable disease by RECIST 1.1, ECOG performance status ≤2, and adequate hematologic, renal, and hepatic function. Any number of prior therapies was permitted.

The five cohorts were (i) liposarcoma, (ii) non-uterine leiomyosarcoma, (iii) undifferentiated sarcoma, (iv) MPNST, and (v) other sarcomas.

treatment plan

Each participant signed an IRB-approved, protocol-specific informed consent in accordance with federal and institutional guidelines. Trial registration ID: NCT01653028.

Patients were treated with alisertib 50 mg orally twice daily on days 1–7, every 21 days. Cycles were repeated every 21 days so long as the absolute neutrophil count was >1 × 10⁹/l and the platelet count was >50 × 10⁹/l on day 1 of the cycle. For treatment-related toxicity including grade 3 or 4 neutropenia, thrombocytopenia, neutropenic fever, nausea, vomiting, or diarrhea, treatment was held until the toxicity improved to ≤grade 2 and then resumed with a dose reduction to 40 mg (first reduction) or 30 mg (second reduction).

evaluation

Patients were evaluated every cycle for treatment adherence and adverse events. National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) v4.0 was used for adverse event coding [9]. Disease assessments (CT or MRI) occurred every other cycle. After 24 weeks, patients were re-evaluated for response every 12 weeks. Response was evaluated using Response Evaluation Criteria in Solid Tumors (RECIST) guidelines (version 1.1) [10]. Upon progression of disease (PD), patients were monitored for long-term adverse events, new primaries, and survival.

correlative studies

Patients treated at Memorial Sloan Kettering Cancer Center underwent optional pre- and post-treatment tumor biopsies, collected before treatment start and following the first week of treatment. Tissues were analyzed by western blot staining. Thirty micrograms of protein were fractionated by SDS–PAGE and transferred on to Immobilon PVDF membranes (Millipore). Membranes were blocked with 5% non-fat milk, probed with primary antibodies and visualized by enhanced chemoluminescence reagents (both from GE Healthcare UK Limited). Antibodies used in this study included rabbit polyclonal antibodies to Aurora A, Aurora B, and phospho-Histone H3 (S10) were purchased from Cell Signaling (#9701, Danvers, MA). Mouse monoclonal antibodies to p53 and to Rb were purchased from BD Parmingen (San Jose, CA) and from Santa Cruz Biotechnology (Dallas, TX), respectively. Rabbit polyclonal for tubulin (Cell Signaling) was used to confirm equal protein loading. These analyses were exploratory and hypothesis-generating.

Select patients also underwent optional FLT-PET scans at baseline and after the first week of treatment. [¹⁸F] FLT is an investigational PET radiotracer that can be used to assess tumor cell proliferation [11]. Since alisertib is a potent inhibitor of AURKA and induces a mitotic arrest, FLT-PET may be a predictive marker of response [12]. FLT was injected and PET emission images were obtained starting ∼60 min after FLT injection.

statistical analysis

The primary end point was confirmed response rate, defined as a CR or PR on two consecutive evaluations at least 6 weeks apart. The study used a Simon two-stage design [12]. A response rate of 5% was considered not promising, whereas a response rate of 25% was considered promising in this population. One response within the first 9 patients in each cohort would expand the cohort to 24 patients. If 3 of 24 patients had a response in a cohort, the result would be positive. This design yielded 90% power to detect a response rate of at least 25% at 0.10 level of significance if the true rate is at most 5%. Confidence intervals (CIs) were calculated using the method of Duffy–Santner [13]. There was a 63% chance of stopping early within a cohort, if the response rate was 5%.

PFS was calculated as the time between registration and the earliest of PD or death. Patients lost to follow-up were censored at their last disease assessment. OS was calculated as the time between registration and death. Patients lost to follow-up were censored at their most recent date last known alive. The Kaplan–Meier [14] method was used to estimate PFS and OS. All patients were followed until the earliest of either death or 18 months post-registration.

Summary statistics and frequency tables are used to summarize baseline patient characteristics and adverse events. All attributions collected for adverse events are reported and are summarized as a maximum severity per patient and type. All data collection and analyses were conducted by the Alliance Statistics and Data Center using SAS version 9.3 and R-2. The dataset was frozen on 18 January 2016. Data quality was ensured by review of data by the Alliance Statistics and Data Center and by the study chairperson following Alliance policies.

results

From August 2012 to December 2013, 72 eligible patients were accrued to this study (liposarcoma: 12, leiomyosarcoma: 10, undifferentiated sarcoma: 11, MPNST: 10, other sarcomas: 29). Table 1 contains a summary of patient characteristics. Thirty-nine (54%) patients were male. The median age was 54.5 (range 20–84).

Table 1.

Patient characteristics (n = 72)

Characteristic	Frequency (%)
Age, median (range)	54.5 (20–84)
Male	54%
ECOG PS
0	42 (58%)
1	27 (38%)
2	3 (4%)
Histology
Liposarcomaa	12 (17%)
Leiomyosarcoma (non-uterine)	10 (14%)
Undifferentiated sarcomab	11 (15%)
Malignant peripheral nerve sheath tumor	10 (14%)
Otherc	29 (40%)

^aThis cohort includes well-differentiated/dedifferentiated liposarcoma (11) and myxoid/round cell liposarcoma (1).

^bThis cohort includes pleomorphic undifferentiated sarcoma (6), MFH and myxofibrosarcoma.

^cPatients enrolled included: osteosarcoma (7), chondrosarcoma (6), synovial sarcoma (3), angiosarcoma (2), fibrosarcoma (2) and 1 each rhabdomyosarcoma, alveolar soft part sarcoma, extraskeletal myxoid chondrosarcoma, endometrial stromal sarcoma, epithelioid sarcoma, Ewing sarcoma, desmoplastic small round cell tumor (DSRCT), and gastrointestinal stromal tumor (GIST).

treatment administration

All 72 patients were treated, receiving a total of 385 cycles (median: 4, range 1–38). Twenty-three treatment delays (range 0–8) were reported for 13 patients and 15 patients required an overall total of 22 dose reductions. All patients have completed study treatment due to: disease progression (83%), adverse events (8%), alternative therapy (1.5%), withdrawal after beginning treatment (1.5%), and other (6%).

toxicity

All patients were evaluable for toxicity. Table 2 summarizes adverse events classified as at least possibly treatment-related for all patients. Common grade 3 adverse events included anemia, neutropenia, thrombocytopenia, leukopenia, febrile neutropenia, oral mucositis, and palmar-plantar erythrodysesthesia syndrome. Grade 4 adverse events frequently occurring included: neutropenia, thrombocytopenia, leukopenia, and febrile neutropenia. One patient with synovial sarcoma received only 2 days of treatment before clinical deterioration consistent with disease progression and died of respiratory failure unrelated to treatment.

Table 2.

Adverse events (n = 72)

CTCAE v4.0 classificationa	Grade 3		Grade 4
	n	%	n	%
Neutrophil count decreased	15	21	15	21
Anemia	10	14	0	0
White blood cell decreased	10	14	6	8
Mucositis oral	8	11	1	1
Lymphocyte count decreased	7	10	0	0
Platelet count decreased	5	7	5	7
Febrile neutropenia	4	6	3	4
Fatigue	3	4	0	0
Palmar-plantar erythrodysesthesia syndrome	3	4	0	0
Leukocytosis	1	1	0	0
Anal mucositis	1	1	0	0
Diarrhea	1	1	0	0
Nausea	2	3	0	0
Typhlitis	1	1	0	0
Vomiting	1	1	0	0
Lung infection	1	1	0	0
Alanine aminotransferase increased	1	1	0	0
Aspartate aminotransferase increased	1	1	0	0
Skin and subcutaneous tissue disorder Other	1	1	0	0
Hypertension	1	1	0	0
Sepsis	0	0	1	1

^aMaximum severity of AE, per patient, and CTCAE v4.0 classification for events considered at least possibly related to study treatment. Patients may appear in this table more than once.

efficacy

All patients were evaluable for response. There was one confirmed PR in the first stage of cohort 5 (Other sarcomas) which led to expansion of that cohort to second-stage enrollment. All other cohorts closed after first-stage accrual. In total, two confirmed responses (7.4%, 95% CI 0.9% to 24.3%) were observed in cohort 5 (Other sarcomas). Both patients had angiosarcoma and were on treatment for 4.6 and 7 months. There were no confirmed responses in the remaining cohorts. Overall, the response rate was 2.8% (95% CI 0.3% to 9.7%) and the median PFS was 11.7 weeks (95% CI 7.0–13). The median PFS (95% CI) within each cohort was: 13 weeks (6.3–37.1) for liposarcoma, 11.7 weeks (1.7–21.9) for leiomyosarcoma, 11.7 weeks (5–20.6) for undifferentiated sarcoma, 13 weeks (3.6–45) for MPNST, and 6.6 weeks (5.9–18.1) for the other sarcomas. The proportion of patients who were progression free at 12 weeks was 73% (38%–91%) for liposarcoma, 44% (14%–72%) for leiomyosarcoma, 36% (11%–63%) for undifferentiated sarcoma, 60% (25%–83%) for MPNST, and 38% (20%–55%) for other sarcomas. The median OS was 68 weeks (23–NR) for liposarcoma, 72 weeks (16–NR) for leiomyosarcoma, 68 weeks (19–NR) for undifferentiated sarcoma, 69 weeks (16–NR) for MPNST, and 29 weeks (17–51) for other sarcomas.

A few patients remained on study with stable disease for a prolonged period. These include one patient in cohort 3 with undifferentiated sarcoma metastatic to the liver and lung on study for 24+ months and three patients in cohort 5 with ASPS, chondrosarcoma, and Ewing sarcoma on study 19, 20, and 26 months, respectively.

correlative studies

Three patients underwent pre- and post-treatment FLT-PET scans. The patients had dedifferentiated liposarcoma, extraskeletal myxoid chondrosarcoma, and leiomyosarcoma. There was no significant change in SUV uptake for each patient comparing the pre- and post-treatment scan.

Another four patients underwent optional paired tumor biopsies. These results are shown in Figure 2. Tissues were analyzed for markers of AURKA and AURKB inhibition. Several of these matched pairs were informative. For example in patient 4, there was evidence of suppression of phospho-H3 (S10). This was associated with decreases in Aurora A, Aurora B, and Rb expression. The inhibition of Aurora A should induce an increase in phosphoso-H3 (S10), a marker of mitotic accumulation. In contrast, phosphos-H3 (S10) is a substrate for Aurora B and an Aurora B inhibitor should inhibit phospho-H3 (S10) [13]. In addition, inhibition of Aurora A should cause an increase in expression of aurora kinases, especially Aurora A, as cells accumulate in mitosis. In patients 2–4, where equal protein loading is confirmed, there is actually loss of Aurora A expression. These events are more consistent with inhibition of Aurora B where the cells are known to undergo polyploidy rather than mitotic arrest. The effect of the drug on Aurora B is also suggested by the observed decreases in Rb as Aurora B (rather than Aurora A) inhibition causes decreases in pRB expression. Unfortunately, all patients experienced PD and received treatment for <12 weeks, thus no association between biomarkers and clinical benefit can be made.

Figure 2.

Western blot analysis of tumor biopsies obtained pre- and post-treatment with alisertib. The four patients had osteosarcoma (1), dedifferentiated liposarcoma (2), and leiomyosarcoma (3 and 4).

discussion

Alisertib was generally well tolerated in patients with advanced sarcoma. The adverse events, as expected, were mostly myelosuppression and were manageable. Alisertib has previously been shown to have activity in several different solid tumors, including breast cancer, lung cancer, head and neck squamous cell carcinoma, gastroesophageal adenocarcinoma, and ovarian carcinoma [12, 14]. In this study, occasional responses were also observed, however not enough for the study to meet its primary end point of overall response. There were however two patients with angiosarcoma who had objective responses, one with angiosarcoma of the scalp and another with radiation-associated angiosarcoma of the breast. The angiosarcoma patients were grouped into the ‘other sarcoma’ cohort and since there were no other histologies that responded, this was not sufficient to meet the response end point. Nevertheless, further study of aurora kinase inhibitors in angiosarcoma would be warranted, especially since angiosarcoma is known to be sensitive to mitotic inhibitors such as taxanes [15].

Other hints of activity were noted in the prolonged stable disease achieved by some patients with ASPS, chondrosarcoma, and Ewing sarcoma. To distinguish whether this is an effect of the drug rather than a reflection of the sometimes indolent natural history of certain sarcoma subtypes would require a randomized trial.

One strength of this study is that patients with different histologic subtypes were enrolled into separate cohorts, rather than combined. This strategy is essential to the success of signal-finding studies since sarcomas are such a heterogeneous group of diseases. Nevertheless, the largest and most successful group was the ‘other sarcoma’ cohort which included a heterogeneous group of different sarcoma subtypes.

The study also confirmed the feasibility and efficiency of doing large phase II studies of rare disease through the Alliance for Clinical Trials in Oncology. The accrual rate to the trial exceeded expectations and demonstrated the power of the cooperative groups [now National Clinical Trials Network (NCTN)] in accruing to sarcoma trials. The other strength is that utilizing the structure of the cooperative groups provides a backdrop of ‘real-world’ care in evaluating new treatments, and capitalizes on the infrastructure of the groups for site efficiency in opening trials, treating patients, and entering data. A potential weakness of the study is that central pathology review was not mandated. Future NCTN studies would be strengthened by having mandatory central pathology review by an expert sarcoma pathologist. Prior data have indicated that can be substantial discordance between local and central pathology review in sarcoma cases, so this is a critical element for future trials [16–18].

Another strength of this investigation is that correlative studies included tumor biopsies done while on-treatment. Although research biopsies were obtained on only a few patients, they did confirm pharmacodynamic effects of the drug. These effects though suggest inhibition of Aurora B rather than Aurora A, which consistent with pre-clinical data that show dose-dependent effects on both [9].

Finally, this study raises the difficult question of optimal end points in sarcoma clinical trials. This was a non-randomized study with a response rate end point. However, many sarcoma experts favor using PFS as the optimal study end point. PFS at 12 weeks of >40% has been considered a sign of an active drug in sarcoma [19]. By this metric, the results of alisertib are promising, especially in liposarcoma (PFS at 12 weeks of 73%), leiomyosarcoma (44%), and MPNST (60%), although the number of patients in each cohort remains small. In comparison, the phase II study of pazopanib demonstrated a response rate of only 6% and PFS at 12 weeks of 26%–49%, depending on the sarcoma subtype. Pazopanib went on to FDA approval based on a phase III study that demonstrated improved median PFS compared with placebo (4.6 versus 1.6 months) yet a response rate of just 6% [3]. Since the alisertib study began, two additional drugs have been FDA-approved for sarcoma: trabectedin (for liposarcoma and leiomyosarcoma) and eribulin (for liposarcoma). Both were compared with dacarbazine in randomized phase III trials and were found to have improved median PFS (4.2 versus 1.5 months for trabectedin; 2.9 versus 1.7 months for eribulin) yet low response rates (10% for trabectedin and 4% for eribulin) [20, 21]. Thus, in this context, the hypothesis that alisertib could result in a 25% response rate seems a high bar. Future studies of new drugs in sarcoma should consider using PFS as the primary end point, ideally in a randomized design.

funding

Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Numbers U10CA031946, U10CA033601, U10CA180821 and U10CA180882 (to the Alliance for Clinical Trials in Oncology), U10CA077597, U10CA180791, and 5U10CA180833, UG1CA189960. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Also supported in part by funds from the Cancer Center Support Grant (P30 CA008748). See supplementary appendix for further details.

disclosure

The authors have declared no conflicts of interest.