OER-073: A Multi-Center Phase II Study Evaluating the Role of Pazopanib in Angiosarcoma

Abstract

Background:

Angiosarcomas are rare mesenchymal sarcomas that can present as primary cutaneous or non-cutaneous disease. They express a variety of vascular endothelial growth factor receptors (VEGFRs). We hypothesized that treatment with pazopanib, a multi-kinase inhibitor with activity against VEGFR, would result in disease response and prolonged disease stabilization.

Methods:

This was an open label phase II trial of pazopanib in incurable angiosarcoma with co-primary endpoints of RECIST response and PFS at 3 months. The starting dose of pazopanib was 800 mg daily.

Results:

Twenty-nine patients were accrued between 2011 and 2018; 22 patients were evaluable for response. Toxicities were similar to those found in prior reports. There was one PR (3%) with a clinical benefit rate (CBR) of (CR + PR + SD) of 48%, observed more frequently in patients with cutaneous disease. The median PFS was 14.4 weeks and the three-month PFS rate determined by Kaplan-Meier estimate was 54.6% (95%CI 36.0, 82.9%), meeting the primary study objective. The Kaplan-Meier estimate of OS was 16.1 months

Conclusions:

Pazopanib therapy in incurable angiosarcoma is associated with meaningful disease control, especially in patients with cutaneous disease with limited objective responses.

Lay Summary:

Angiosarcoma is a rare cancer. It can be found on the skin or in internal organs. This study tested pazopanib, an oral targeted medication, to determine its benefit in patients with angiosarcoma that could not have their tumors removed by surgery. Pazopanib treatment was safe and no new side effects were reported. The study showed that pazopanib controlled tumor growth in half of patients at 3 months, and was more common in angiosarcomas of the skin; it led to tumor shrinkage in a minority of patients (1 out of 29).

Precis for Table of Contents:

Pazopanib therapy in incurable angiosarcoma is associated with meaningful disease control, especially in patients with cutaneous disease with limited objective responses. Toxicity in this patient population is consistent with prior reports.

Introduction

Angiosarcomas are rare and aggressive mesenchymal cancers that arise from endothelial cells. They account for about 2% of all soft tissue sarcomas (STS) [1, 2]. The incidence of angiosarcomas has risen over the past 30 years, but whether this is an actual increase remains unclear. The disease can present at any age but is frequently seen in the elderly. The skin is the primary site of disease in about half of the cases. These sarcomas can also arise in deep structures, with about 10% of cases involving viscera. Most angiosarcomas arise spontaneously. Chronic lymphedema, radiation therapy, exogenous toxins, foreign bodies, and genetic diseases such as Neurofibromatosis-1, mutated BRCA1 or BRCA2, Maffucci syndrome, and Klippel-Trenaunay syndrome have been reported as etiologic factors [3]. The prognosis varies depending on the site of origin. The 5-year survival has been reported to be 30%–40%, with poorer survival seen in visceral disease, older age, retroperitoneal location, large size, and high Ki-67 values [4–6].

Therapy for angiosarcomas remains challenging. Surgery is curative therapy for localized disease, and is often combined with neoadjuvant and/or adjuvant radiation and systemic therapy in a multidisciplinary fashion [7, 8]. Angiosarcomas have a high propensity for recurrence and metastasis, eventually managed with systemic therapy. Few systemic options exist, and response rates to existing agents are poor. Additionally, given the rarity of the disease, most studies of STS have included only a limited number of angiosarcoma patients. Older reports of doxorubicin-based regimens tested in soft tissue sarcomas report response rates of 10–33% in angiosarcoma [9–11]. Paclitaxel, in angiosarcoma specific trials, given either weekly or every three weeks, is another active alternative as evidenced by a 74% and 45% PFS at two and four months respectively [12,13].

Angiosarcomas express upregulation of several vascular-specific receptor tyrosine kinases, including VEGFR-1/−2, and TIE1/2 [14]. In addition, radiation therapy associated secondary angiosarcomas are characterized by amplification of MYC and FLT4, which encodes VEGFR3 [15]. Preclinical studies in angiosarcoma xenograft models have demonstrated antitumor effects of antiangiogenic agents [16]. As a result of these findings, agents that target the VEGF pathway have been of interest for treatment of angiosarcomas.

Clinical studies evaluating anti-angiogenic agents, including sorafenib and bevacizumab, have shown encouraging single-agent activity in this disease [17, 18]. Pazopanib is a multi-target, small-molecule inhibitor of VEGF receptor (VEGFR) (−1, −2, and −3), platelet-derived growth factor receptor PDGFR (-α and -β), and c-Kit. This drug is effective and safe in the treatment of several malignancies, and was FDA approved for advanced STS in 2012. We hypothesized that pazopanib, with its anti-angiogenic mechanism, would have therapeutic activity in angiosarcoma. We report on the activity of pazopanib in angiosarcomas.

Methods

Study design

This was a phase II, open-label, multi-institutional, single-arm study of pazopanib for the treatment of patients with incurable angiosarcoma.

Study Population

Eligible patients were age 18 years and older. All patients had a pathologic diagnosis of advanced stage angiosarcoma not amenable to treatment with curative intent. Patients needed to have an Eastern Cooperative Oncology Group (ECOG) performance status of ≤ 2, measurable disease per RECIST v.1.1 or cutaneous disease amenable to serial measurements, as well as adequate hematologic, renal and hepatic function as assessed by lab testing.

Notable exclusion criteria included history or clinical evidence of central nervous system (CNS) disease including metastases, leptomeningeal involvement, or cerebrovascular accident or transient ischemic attack; abnormalities associated with an increased risk for bleeding; cardiac disease including prolonged corrected QT interval (QTc) > 480 msecs using Bazett’s formula, LVEF < 50%, class III or IV congestive heart failure, or poorly controlled hypertension; thromboembolic disease within the previous six months; and known endobronchial lesions and/or lesions infiltrating major pulmonary vessels. A wash out of 14 days or five half-lives of a drug (whichever was longer) prior to the first dose of pazopanib was required. Prior exposure to pazopanib or a VEGFR targeted kinase therapy, except for bevacizumab or VEGFR-Trap, was not allowed.

Treatment

Pazopanib was taken orally without food at least one hour before or two hours after a meal at a daily dose of 800 mg continuously; cycles were 21 days in length. Subjects received study treatment until disease progression, intercurrent illness that prevented further treatment, unacceptable adverse events, treatment interruptions > 28 days, ≥ 2 dose modifications, patient withdrawal from the study, or changes in the patient’s condition that rendered the patient unacceptable for further treatment in the judgment of the treating investigator.

The National Cancer Institute Common Terminology Criteria for Adverse Events Version 4.0 (NCI-CTCAE, v.4) was used to grade the severity of adverse events (AEs). Two-dose reductions were permitted in a stepwise fashion (600 mg PO daily and subsequently to 400 mg PO daily). Pazopanib was discontinued for at least 7 days prior to surgery. Following surgery, adequate wound healing was determined by the treating physician prior to medication re-initiation. After discontinuation of the study treatment, subjects were followed for disease status and subsequent anti-cancer treatments until death or for a maximum of two years from the date of the last subject enrolled. Patients discontinuing study treatment for unacceptable study related adverse events were followed until resolution or stabilization of the adverse event.

Study End Points

The co-primary endpoints were progression free survival (PFS) at 3 months and objective response rate defined as complete response (CR) and partial response (PR) in angiosarcoma patients treated with pazopanib. The secondary endpoints were the assessment of overall survival of patients treated with pazopanib and safety data for pazopanib in this patient population.

Statistics

The planned sample size was 30 patients. The study power, type I error and chance of early termination were determined under four end-point conditions: (null hypothesis: response=5%, PFS at 3 months=5%), (response=5%, PFS at 3 months=20%), (response=20%, PFS at 3 months=5%) and (response=20%, PFS at 3 months=20%). Since study power and chance of early termination depend on the relation of the two endpoints, we provided these four different scenarios: 1) independence of response and PFS at 3 months, and 2) fullest dependence of these endpoints, which could occur in two ways: 1) p(response and pfs) = min(pr,ps), and 2) p(response and pfs)= max(0,pr+ps-1) where pr=p(response) and ps=p(pfs).

Using a double end-point design with n1=15, n2=30, 15 patients were initially enrolled and evaluated for PFS at 3 months and response at any time up to 3 months. If at least 2/15 patients responded or at least 2/15 patients were progression free at 3 months, then another 15 patients would be recruited. Otherwise, the study would be terminated after the initial 15 patients were evaluated and the null hypothesis accepted. If at least 4/30 patients responded or at least 4/30 patients were progression free at 3 months, we would reject the null hypothesis and accept the alternative that either the response rate was at least 20% or PFS at 3 months is at least 20%. Otherwise, the null hypothesis would be accepted. The study had 69% chance of early stopping under the null, and 14% under the alternative. Under either the independence or dependence II assumption, the study had an overall power of 81% and 9.2% type I error; however, under the assumption of dependence I, power is decreased to 78.5% while type I error is reduced to 4.5%.

At completion, PFS was estimated using the method of Kaplan and Meier. Patients lost to follow up or followed until the end of the study were censored at the time and conditions of their last visit. Standard estimates of the binomial proportion were used to estimate, and place confidence bounds on the several response rates. Overall survival and toxicity data were collected and summarized in a descriptive fashion.

Study Oversight

The study was conducted after approval by the IRB of the participating institutions. Informed consent was obtained from all patients. The Fox Chase Cancer Center Data Safety Monitoring Committee oversaw the trial.

Results

Patient Characteristics

From November 2011 to December 2018, twenty-nine patients were enrolled at five centers; patient characteristics are summarized in Table 1. The median age of the study population was 67 years with a median performance status of 1. The median time between the diagnosis of angiosarcoma and the inclusion in this trial was 19.2 months. Fifty-nine percent of the population was female with 48% of patients having cutaneous primary tumors. There were 4 (14%) radiation related tumors. Twenty-four patients (83%) had received previous systemic therapy for management of angiosarcoma. Taxane-based regimens were the most common prior therapy, with 23 patients (79%) having received either paclitaxel or docetaxel, of which 19 (83%) received the taxane in the first line setting. Fifty-five percent of patients had received 2 or more prior chemotherapy regimens. There were three patients that had received prior bevacizumab in combination with chemotherapy.

Table 1:

Patient Characteristics

Patient Characteristic	N
Sex : M/F	12/17
Age, median	66
range	30–84
ECOG PS: 0, 1, 2	6, 16, 3
Prior therapy:
Surgery	6
Radiation	7
Chemotherapy	24
Number of prior regimens:
Range: 0–6
Median: 2
Angiogenic therapy	3

Efficacy

Of the 29 patients enrolled on study, 2 did not initiate therapy. The response rates as of the data cutoff date of 1/20/19 are summarized in Table 2. The best response was partial response (PR) in one patient (3%), with stable disease (SD) seen in 13 patients (45%); no complete responses were observed (CR). CBR (CR + PR + SD) was 48%. Stable disease lasting longer than three months was seen in four patients with cutaneous disease and one with visceral disease. Response to therapy was not evaluable in 7 patients: two never started pazopanib (lab abnormalities day 1, n=1; consent withdrawal, n=1); four discontinued prior to disease assessment due to toxicity, and one was non-compliant. The study met the primary endpoint for PFS at 3 months, but did not meet the measurable response endpoint of 20% or higher.

Table 2:

Objective Response in the Intent to Treat Population

Best Response in Intent to Treat Population	N, %
Complete Response	0 (0%)
Partial Response	1 (3%)
Stable Disease	13 (45%)
SD > 3 months	5 (17%)
Progressive Disease	8 (28%)
Not evaluable*	7 (24%)

^*2 patients did not start therapy. 5 were not evaluable as removed from study secondary to toxicity (4) and non-compliance (1).

The three-month PFS rate determined by Kaplan-Meier estimate was 54.6% (95%CI 36.0, 82.9%) (Figure 1). Median PFS was 14.4 weeks, 95% CI = [10.1, upper bound not evaluable]. The median overall survival as determined by Kaplan-Meier estimate was 16.1 months.

Figure 1A and B.

depict Kaplan Meier Estimates of Progression Free and Overall Survival of all patients accrued to the study, including patients that were not eligible for disease response assessment.

Safety

Table 3 summarizes the toxicities observed in patients. Toxicities were similar to those described in prior reports [19, 20]. Grade 3 and 4 toxicities included hypokalemia, increased urine protein/creatinine ratio, lymphopenia, hypertension, and abnormal LFTs; there was one death due to cardiac arrest considered possibly related to pazopanib.

Table 3:

Toxicity

Toxicity (Related/Unrelated in >10%)	Grade 1 / 2 N (%)	Grade 3 / 4 N (%)	Total N (%)
Cardiac: Hypertension	7 (24)	8 (28)	15 (52)
Constitutional: Dizziness	7 (24)	0 (0)	7 (24)
Fatigue	19 (66%)	1 (3)	20 (69)
Anorexia	11 (38)	0 (0)	11 (34)
Dermatologic: Rash, including PPE	7 (24)	0 (0)	7 (24)
Gastrointestinal: Nausea	15 (52)	0 (0)	15 (52)
Vomiting	11 (38)	0 (0)	11 (38)
Diarrhea	13 (45)	2 (7)	15 (52)
Constipation	9 (31)	1 (3)	10 (34)
Hematologic: Decreased WBC	8 (28)	0 (0)	8 (28)
Anemia	9 (31)	1 (3)	10 (34)
Thrombocytopenia	8 (28)	2 (7)	10 (34)
Hepatic: Increased AST and ALT	8 (28)	5 (17)	13 (45)
Increased Alk Phos	6 (21)	1 (3)	7 (24)
Increased bilirubin	7 (24)	1 (3)	8 (28)
Electrolyte Abnormalities: Hyperglycemia	9 (31)	0 (0)	9 (31)
Hyponatremia	4 (14)	0 (0)	4 (14)
Hypocalcemia	4 (14)	0 (0)	4 (14)
Elevated creatinine	5 (17)	0 (0)	5 (17)

Of 100 completed cycles of therapy in 27 patients, 32 cycles required holding pazopanib therapy. The most common reasons for holding therapy were grade 3 hypertension (n=4), grade 2/3 abnormal liver function tests (n=2), and grade 3 gastrointestinal (GI) toxicities, including nausea, diarrhea, or abdominal pain (n=3). The median number of cycles was 2 (range: <1 to 20 cycles); 10 patients received more than 3 cycles of therapy.

There were ten dose modifications from the initial dose of 800 mg daily, six dose reductions to 600 mg daily, four additional dose reductions to 400 mg daily. The reasons for dose reductions included: grade 2 dysgeusia, diarrhea, and anorexia with weight loss, pain (n=1 for each side effect) and palmar plantar erythrodysesthesia syndrome (n=2), and grade 3 hypertension, pain/muscled cramps or liver function tests (n=2 each), as well as increased urine protein to creatinine ratio (n=1). Thirteen patients required a dose to be held for side effect management, range 1–16 days, median 7 days. Eight of the patients needed more than one dose hold, ranging from 2–5, with a median of 3. The mean dose-intensity for patients who completed one cycle or more was 85%.

Discussion

Advanced angiosarcomas are aggressive malignancies with a poor prognosis due to the limited durability and low response rates to conventional systemic therapy. The current standard for the management of incurable angiosarcomas is systemic chemotherapy. Recommended therapies include taxane and anthracycline based regimens, as well as gemcitabine alone or in combination with taxanes. Here we report the results of a phase II, open-label, multi-center, single-arm study of pazopanib in this patient group. The trial had co-primary endpoints, PFS at 3-months and objective response rate. The Kaplan-Meier estimate of PFS at 3 months was 54.6% and met the primary PFS at the 3-month endpoint, while the objective response rate of 3% was less than 20%, and thus did not meet the response endpoint. The CBR was 48%. The Kaplan-Meier estimate of OS was 16.1 months with long term disease control seen mostly in cutaneous angiosarcomas. Additionally, toxicities were manageable and similar to those reported in the medical literature [19, 20].

Several chemotherapeutic agents have been used to treat advanced angiosarcoma. Traditionally soft tissue sarcomas are treated with anthracycline-based chemotherapy [9–11]. The EORTC compared liposomal doxorubicin and doxorubicin in a phase II randomized trial [9]. They found that response rates of 10 and 9 % respectively in advanced soft tissue sarcomas, 12% of which were angiosarcomas. More recent prospective and randomized trials, including only angiosarcoma patients, suggest that taxane based chemotherapy is also an active treatment option for these tumors with response rates ranging from 18 to 45.8%, with a median time to progression of 4 months and OS of about 8–19.5 months; the more favorable outcomes were seen in patients who had received 2 or fewer prior lines of therapy [12,13]. Eribulin, in patients with prior taxane therapy, has a response rate of 20%, like that that seen in more heavily pretreated patients receiving paclitaxel [21]. Trabectedin, in a retrospective analysis, demonstrated a 22% PFS at 3 months, which does not meet the criteria for an effective agent based upon the EORTC 3 month PFS analysis [22]. Clearly, there is an unmet need for additional therapy for patients with incurable angiosarcomas.

Given the biology of angiosarcomas, other agents targeting VEGF/VEFGR have also been tested. Sorafenib was noted to have a 14% response rate [17], while bevacizumab was associated with a 7% response rate [18]. An EORTC retrospective analysis of patients treated with pazopanib, some on clinical trials, reported a response rate of 20%, with similar responses in cutaneous and non-cutaneous disease [23]. The response rate in tumors that were radiation associated was also 20%. The median PFS and OS were 3 and 9.9 months, respectively. Other small case series have also noted similar findings [26, 27]. The outcomes in our study demonstrated lower response rates, but longer PFS and OS, providing additional data to support the use of pazopanib in angiosarcoma, especially after failure of other conventional regimens.

Although there is clinically meaningful activity with pazopanib for treatment of angiosarcomas, additional therapeutic advances are needed. More recently whole genome sequencing studies have been performed [28,29]. Head and neck primary angiosarcomas demonstrated UV mutation signatures as well as increased tumor mutation burden, which would support use of immunotherapy. A retrospective analysis of immunotherapy biomarkers in angiosarcomas also identified greater numbers of head and neck tumors with elevated tumor mutation burden (>10 muts/MB) as well as PD-L1 positivity (noted in >20% of cases from head and neck, visceral, and unknown sites) [30]. In the phase II trial of Dual anti-CTLA-4 and anti-PD-1 Blockade in Rare Tumors (DART) in metastatic or unresectable angiosarcoma, the overall response rate was 25%; of the 4 responses seen, 3 were from the 5 patients with primary scalp or face tumors [31]. Trials combining VEGFR TKI’s and immunotherapy in sarcomas have been reported, but not using pazopanib. Sunitinib with nivolumab was tested in patients with UPS, synovial sarcoma, epithelioid sarcomas and those with known response to antiangiogenic therapies [32]. Upon initiation of nivolumab on day 15, a dose reduction of sunitinib from 37.5 mg daily to 25 mg daily was required. The RECIST 1.1 response rate was 21%. A second study evaluated axitinib with pembrolizumab and enrolled patients that had confirmed sarcomas with evidence of progression within the past 6 months. The RECIST response rate was 25% with the majority of patients that responded having a diagnosis of alveolar soft part sarcoma. Given the benefit seen in cutaneous angiosarcomas in our study, as well as the UV signature noted in head and neck angiosarcomas, further investigation into combination VEGFR-targeted and immunotherapies is warranted.

Our study had several limitations. Angiosarcomas are rare tumors which made it a challenge to accrue. Additionally, the accrual rate was slower than expected possibly because pazopanib became available off-study. Furthermore, response was not evaluable in 7 patients due to failure to start the medication, toxicity, or non-compliance.

In conclusion, we found that treatment with pazopanib was feasible and was not associated with any unreported toxicities. There was no evidence of increased risk of tumors bleeding. Also, the efficacy outcomes were limited by early withdrawal of patients. Lastly, while only one RECIST responses was observed, pazopanib had a clinically meaningful 3-month PFS, primarily in cutaneous AS.