A phase Ib trial of the combination of the anti-angiogenic agent sunitinib and radiation therapy for patients with primary and metastatic central nervous system malignancies

Abstract

Purpose

This phase I trial evaluated sunitinib combined with radiation therapy (RT) for primary or metastatic central nervous system (CNS) malignancies.

Experimental Design

Eligible patients had CNS malignancies requiring a (minimum) two week course of RT. Sunitinib (37.5 mg) was administered daily for the duration of radiotherapy treatment with optional treatment extension of a month. Urine was collected at 3 time points for correlative biomarker studies. The primary endpoint was acute toxicity defined by the Common Toxicity Criteria version 3.

Results

Fifteen patients were enrolled (12 CNS metastasis; 3 primary tumors). RT doses ranged from 14–70 Gy (1.8–3.5 Gy per fraction). Acute toxicities included hematologic, nausea, hyperglycemia, fatigue, hypocalcemia, and diarrhea. Six patients (40%) developed grade ≤2 toxicities. Grade 3 toxicities occurred in 7 (47%) patients including hematologic, fatigue, deep vein thrombosis, dysphasia, hyperglycemia and hyponatremia. No grade 3–5 hypertensive events or intracerebral hemorrhages occurred. Two Grade 5 adverse events attributed to disease progression occurred. Median follow up was 12.4 months. Two (13%) patients had partial response, 9 (60%) patients had stable disease and 2 (13%) patients had progressive disease. Six-month progression-free survival of brain metastasis patients was 58%. Grade 3 hematologic toxicity correlated with higher VEGF changes between baseline and completion of RT.

Conclusion

Continuous 37.5 mg sunitinib combined with RT in CNS malignancies yields acceptable toxicities and adverse events. Changes in urine VEGF were associated with hematologic toxicity and should be analyzed in a larger cohort. The feasibility, safety, and early response results warrant a Phase II trial.

Purpose

Preclinical studies suggest that the combination of angiogenic blockade and RT can enhance the therapeutic ratio of ionizing radiation via the phenomenon of vascular normalization^{1, 2}. Though counterintuitive and poorly understood, combining ionizing radiation and anti-angiogenic agents may cause aberrant tumor vessels to regress^{3, 4} thereby increasing tumor oxygen concentration and acting as a potential radiosensitizing agent. Anti-angiogenic agents may also decrease interstitial fluid pressure within tumors allowing improved oxygen delivery to hypoxic tissues^{5, 6}.

Further, anti-angiogenic agents have multiple other effects on the tumor microenvironment which may impact the tumor response to radiation therapy. Agents which disrupt the phosphoinositol-3 kinase (PI3K)-Akt-mTOR signaling pathway may radiosensitize the vascular endothelium⁷. The ability of endothelial progenitor stem cells which contribute to tumor angiogenesis and growth to mobilize and recruit can be impaired by agents that disrupt the PI3K-Akt-mTOR pathway^8–10.

Preclinical studies in glioma¹¹ and meningioma¹² suggest anti-angiogenic agents may have an enhanced radiosensitizing effect in central nervous system (CNS) tumors.

SU11248 (sunitinib) is an orally active, indolinone-based, multitargeted receptor tyrosine kinase inhibitor that selectively targets vascular endothelial growth factor receptors (VEGFR1, VEGFR2, VEGFR3), platelet-derived growth factor receptors (PDGFRα, PDGFRβ), stem cell factor (c-KIT), neurotropic factor receptor (RET), FMS-like tyrosine kinase-3 (FLT3), and colony stimulating factor (CSF-1R)^13–22. Sunitinib is FDA approved as monotherapy in metastatic renal cell carcinoma and imatinib-refractory gastrointestinal stromal tumor (GIST).

Despite an array of tumor markers currently in use, none serve as general predictors of outcome after therapy. In theory, angiogenic factors can identify patients at risk for recurrent disease, regardless of tumor type since the process of angiogenesis is ubiquitous to cancer. Urine biomarkers provide a non-invasive platform to interrogate disease status and tumor biology. Based on this information we hypothesized that urine VEGF measured at baseline, at the end of radiation, and at one month post-radiation may be predictive of treatment response.

The combination of RT and VEGF pathway targeting anti-angiogenic agents have shown tumor growth delay and tumor cell killing in pre-clinical studies^23–27. In addition, preclinical models with sunitinib and ionizing radiation have demonstrated a more than additive effect on tumor growth delay compared with either treatment alone^{3, 28, 29}.

We evaluated the safety and toxicity profile of 37.5 mg of oral sunitinib administered daily to patients undergoing RT for primary or metastatic central nervous system malignancies. In addition, we evaluated correlative urine biomarker results and early response rates for this regimen.

Experimental Design

Patient Selection

This study received approval of the Kimmel Cancer Center Research Review Committee (CCRRC) and Thomas Jefferson University Institutional Review Board (IRB) (IRB #06C.549) prior to patient recruitment and accrual. Patients with histologically confirmed primary or metastatic CNS malignancies whose treatment involved a minimum 2-week course of RT were eligible. Patients with widely metastatic disease with brain metastases did not have biopsy confirmation of the brain metastases in situations where MRI appearance was characteristic of brain metastases. Other eligibility requirements included: age ≥18 years, World Health Organization (WHO) performance status of 0–2³⁰, and life expectancy >3 months. Prior surgery, chemotherapy and/or RT (including radiation to the CNS) was allowed provided there was resolution of all acute toxic effects of prior therapies (less than or equal to grade 1 by the National Cancer Institute Common Toxicity Criteria [CTCv3])³¹. Adequate hematologic, hepatic, and renal function (defined as hemoglobin >9 g/dL, platelets >100,000, absolute neutrophil count >1500/µL, AST & ALT <2.5 times upper limit of normal [ULN], total serum bilirubin <1.5 times ULN, serum calcium <12 mg/dL, and serum creatinine <1.5 time ULN) was required.

Study Design

The study involved a baseline screening assessment period, treatment, and an observation period. During treatment, sunitinib 37.5 mg was given orally starting the first day of RT and was taken daily throughout the RT course including weekends. This regimen was chosen based on multiple phase I and II studies in metastatic renal cell carcinoma^{32, 33}, GIST³⁴, and soft tissue sarcoma³⁵ which demonstrated this schedule was well tolerated with a manageable toxicity profile. In addition, a continuous daily regimen was hypothesized to maximize the potential radiosensitizing effect. Total dose and fractionation of RT varied depending on tumor type, prior RT doses, and physician preference. Dose limiting toxicities (DLT) were defined as any grade 3–5 toxicity attributable to sunitinib, excluding grade 3 hematologic toxicity unless clearly dose limiting. Early termination of the study was stipulated if the DLT rate exceeded 50%. A provision for de-escalation to sunitinib 25 mg using the standard 3+3 phase I trial design (Design 1 described by Simon et al.³⁶) was permitted if more than one third of patients enrolled experienced a DLT. Patients who experienced minimal toxicity were given the option to continue sunitinib for an additional 30 days after completion of RT. All patients had weekly hematology and chemistry panels. Corticosteroid use prior to treatment, during treatment, and at one month followup was documented. Corticosteroids were not required, but were used as necessary at the discretion of the treating physicians.

The primary endpoint of this study was to evaluate the toxicity and safety profile of combining sunitinib and RT. Secondary endpoints included assessment of radiographic tumor response at one month and urine biomarker changes.

Toxicity Grading and Evaluation of Response

All toxicities were documented according to CTCv3 criteria³¹, and patients were evaluated for potential DLT weekly during treatment and 4 weeks after completion of treatment. Pertinent history, physical exam, hematology and chemistry studies were completed weekly while on treatment and one month after completion of treatment. MRI or CT scan were completed approximately one month after study completion and at regular follow-up appointments thereafter, dependent on the diagnosis, to assess early response. Formal neuro-cognitive functional assessments were not performed. Early imaging response was assessed using the Macdonald criteria³⁷.

Patients, urine collection and processing

Urine was collected from all study patients at baseline, completion of RT, and 1 month post treatment. Urine was stored at −20 °C prior to evaluation, then centrifuged at 3000 rpm for 10 min at 4 ˚C. The supernatant was used for subsequent assays. Urinary creatinine levels were obtained on the Bayer DCA 2000+ Analyzer (Bayer HealthCare, Elkhart, IN) according to the manufacturer’s protocol. All samples were normalized by creatinine level. Urine samples were assayed for VEGF, MET, MMP-2, and MMP-9. These biomarkers were chosen for analysis because of prior published data suggesting an association between these biomarkers and brain tumors^{38, 39}.

Electrochemiluminescence immunoassays

MET, MMP-2, MMP-9, and VEGF were measured in urine specimens via electrochemiluminescent immunoassay (Meso Scale Discovery, Gaithersburg, MD). All samples were thawed at room temperature for 4 hrs. prior to evaluation. MMP-2, MMP-9 and VEGF assays were performed per the manufacturer’s protocol. MET levels were detected according to a previously published protocol⁴⁰.

Statistical Methods

To assess clinical efficacy, the neurologic progression–free survival (PFS) rate and overall survival (OS) rate were calculated from the time of enrollment using the Kaplan–Meier method⁴¹. Median urinary VEGF levels over time were compared using Friedman’s chi square test based on ranks. Median levels of VEGF fold-change were compared based on the development of hematological toxicity using the exact Wilcoxon rank sum test. All statistical tests are 2-sided and p-values less than or equal to 0.05 are considered significant.

Results

Patient characteristics and treatment

Fifteen patients were accrued from April 16, 2007- January 7, 2008. Table 1 displays patient characteristics and total radiation treatment dose and fraction size. Two patients had been previously treated with radiation therapy to the brain. The median total dose was 37.5 Gy (range, 14–70 Gy) including five patients treated to >50 Gy. The median dose per fraction was 2.5 Gy (range, 1.8–3.5 Gy). Thirteen patients completed the prescribed treatment. One patient expired during treatment and another patient chose to discontinue sunitinib secondary to decreased blood counts, vomiting, and fatigue. Seven (47%) patients chose to continue daily sunitinib for 30 days after completion of RT.

Table 1.

Patient and treatment characteristics

	All patients	Metastatic Disease	Primary Tumor
No. Patients	15	12	3
Histology
Melanoma	4	4	0
NSCLC	3	3	0
Adenoid cystic	2	2	0
HNSCC	1	1	0
Chordoma	1	1	0
Anaplastic meningioma	1	0	1
CNS sarcoma	1	0	1
WHO grade II astrocytoma	1	0	1
Prior RT	2	2	0
Age (Median)	58.7 (31–77)	54.4 (37–77)	50.7 (31–76)
<45	2	1	1
45–54	3	2	1
55–64	5	5	0
65–74	3	3	0
75+	2	1	1
Gender
Male	12	9	3
Female	3	3	0
Race
Caucasian	10	9	1
African-American	3	3	0
Asian-American	2	0	2
WHO Performance
0	3	2	1
1	12	10	2
Target
Whole brain	9	9	0
Partial brain	6	3	3
RT Dose (median, Gy)	37.5 (14–70)	37.5 (14–59.4)	60.0 (50.4–70)
RT dose per fraction
1.8 Gy	2	1	1
2.0 Gy	2	0	2
2.5 Gy	8	8	0
3.0 Gy	1	1	0
3.5 Gy	2	2	0
Total Dose
<25 Gy	1	1	0
25–30 Gy	2	2	0
35–45 Gy	7	7	0
>45 Gy	5	2	3
Median weeks of sunitinib treatment (range)	5 (0.6–10.8)	4.7 (0.6–10.8)	6.4 (2.6–8.8)

Toxicity Evaluation

Six patients (40%) had grade 1 or 2 toxicities including fatigue, hyperglycemia, nausea, hypocalcemia, and diarrhea. Hematologic toxicities were most common with nearly all patients having at least grade 1 anemia, leukopenia, or thrombocytopenia. Toxicities by grade are listed in Table 2.

Table 2.

Toxicity Profile

Toxicities	Grade 1	Grade 2	Grade 3	Grade 4	Grade 5	Any
Hematologic
Leukopenia	2	2	2			6
Thrombocytopenia	3	1	1			5
Anemia	2	2				4
Lymphopenia			1			1
Neutropenia			1			1
Metabolic
Hyponatremia	3		1			4
Hyperglycemia	3		1			4
Hypocalcemia	1	1				2
Hypercarbia	1					1
Hyperuremia	1					1
Hypokalemia	1					1
Elevated creatinine	1					1
Hepatic
Hyperbilirubinemia	1	1				2
Hypoproteinemia	1					1
Hypoalbuminemia	1					1
Elevated Alk. Phosphatase	1					1
Elevated ALT	1					1
Elevated AST	1					1
Constitutional
Fatigue	3	3	2			8
Pain		2				2
Dermatitis	1	1				2
Chest Pain	1	1				2
Alopecia	1	1				2
Edema	1					1
Gastrointestinal
Nausea	3	1				4
Anorexia	3					3
Diarrhea	1	1				2
Dyspepsia	1	1				2
Emesis		1				1
Mucositis		1				1
Neurologic
Seizure		1			1	2
Dysphasia			1			1
Drooling/Difficulty chewing			1			1
Headache	1					1
Motor Neuropathy	1					1
Vascular
DVT		1	1			2
Pulmonary Embolism					1	1
Dyspnea	1					1
Epistaxis		1				1
Vaginal bleeding	1					1
Hypertension	0					0
All	43	23	12	0	2	80

Grade 3 toxicities occurred in 7 (47%) patients. Two (13%) grade 3 toxicities met the criteria for DLT. The other grade 3 toxicities were thought not to be attributable to the combination of sunitinib with RT and were not dose limiting. Of the twelve grade 3 toxicities, 7 were thought to be attributed to sunitinib (including neutropenia, leukopenia, thrombocytopenia, hyponatremia, and hyperglycemia), 2 were attributed to RT (dysphasia, difficulty chewing), two were attributed to the combination (fatigue) and one DVT was attributed to the prevalent coagulation abnormalities among patients with brain tumors. No difference in toxicity was observed between patients treated with whole brain RT (WBRT) versus partial brain RT. No difference in toxicity was observed based on radiation therapy fraction size or total dose administered. There was no observed difference in skin toxicity or toxicity of other tissues within the radiation portal. No grade 4 or 5 hypertensive events or intracerebral hemorrhages occurred. Two Grade 5 adverse events (status epilepticus and pulmonary embolism) occurred and were determined to be related to disease progression by the medical monitor. One patient with metastatic melanoma who was retreated with WBRT after prior WBRT and stereotactic radiosurgery died on day 4 of treatment from uncontrolled status epilepticus attributed to multiple brain metastases. A second patient with multiple brain metastases from non-small cell lung cancer died six days after study completion from a presumed pulmonary embolism.

Early Response

Thirteen patients had follow up MRI imaging to assess one-month tumor response. Two (15%) patients demonstrated a partial response, nine (70%) patients had stable disease and 2 (15%) patients had disease progression. All three patients with primary CNS tumors had stable disease. Among patients with metastatic disease, 2 (20%) had partial response, 6 (60%) had stable disease, and 2 (20%) had progressive disease (Table 3). Prior to initiation of treatment, 5 (33%) patients were on high-dose corticosteroids. At 1-month follow up none of the patients required corticosteroids. Three (20%) patients initiated corticosteroid therapy during RT, and all were tapered off corticosteroids within one month after RT.

Table 3.

Early treatment response as determined by gadolinium-enhanced MRI 1 month post treatment and 6-month PFS and median overall survival

	All patients	Metastatic Disease	CNS Primary
Partial Response	2 (15%)	2 (20%)	0 (0%)
Stable Disease	9 (70%)	6 (60%)	3 (100%)
Progressive Disease	2 (15%)	2 (20%)	0 (0%)
6-month PFS	9 (60%)	8 (67%)	1 (33%)
Median OS	8.8 months	7.6 months	Not reached

One patient initially presented with multiple small parenchymal metastatic brain lesions from adenoid cystic carcinoma, the largest was in the right parietal lobe and measured 14 × 10 mm with an additional 12 × 8 mm lesion in the right frontal lobe. After finishing RT and continuing sunitinib for one month, an MRI revealed an interval decrease in the size of all the previous parenchymal enhancing lesions. Another patient with metastatic renal cell carcinoma presented with a 4.6 cm dural based mass involving the sphenoid sinus, sella, suprasellar cistern, clivus and cribriform plate with residual tumor after transphenoidal resection. One month follow up MRI revealed a significant decrease in the size of the skull base lesion with no enhancement and no metabolic activity on PET.

Survival

With a median follow up of 34.2 months, 6 patients continue to have stable disease and 7 have disease progression. PFS at six months was 60%, with 9 patients having stable disease (Figure 1). Four of the original 15 (27%) patients remain alive. Median survival is 8.8 months (Table 4). Eleven (73%) patients were alive six months after treatment, with a survival range for all patients of 0.1 to 39.9 months. Of the patients with brain metastases 2 of the 12 patients (18%) were alive (range 0.1 to 39.9 months). Seven (58%) of the patients with brain metastases were free of progression for 6 months.

Figure 1.

Kaplan-Meier overall survival and neurologic progression-free survival

Table 4.

Histology, response, neurologic progression-free survival and overall survival

Patient	Histology	Response	Neurologic PFS	Overall Survival (months)
1	Adenoid cystic	PD	7.4	7.4
2	Non-small cell lung	NA	0.7	0.7
3	Melanoma	NA	0.4	0.4
4	Astrocytoma	SD	19.8	20.9
5	Non-small cell lung	SD	4.9	4.9
6	HNSCC	SD	15.7	39.9
7	Chordoma	SD	7.8	7.8
8	Atypical Meningioma	SD	11.5	11.5
9	Non-small cell lung	SD	28.9	28.9
10	Melanoma	SD	20.8	20.8
11	Melanoma	SD	3.2	6.0
12	Adenoid cystic	PR	6.0	8.8
13	Melanoma	PD	4.3	5.4
14	Renal cell	PR	18.3	20.0
15	CNS sarcoma	SD	7.8	16.7

Urine VEGF

Urine samples were collected from 13 of the 15 patients enrolled on this study. Samples at baseline, end of radiation, and one month post-radiation were collected for 12/13, 10/13, and 5/13 patients, respectively. Urine VEGF was detectable in 24/27 (89%) samples. The median VEGF concentration for all time points for patients with detectable levels was 336 pg/mg (range 53–1153 pg/mg). There were no statistically significant differences in VEGF levels between any of the three time points, see Figure 2 (Friedman’s chi square test, p=0.87).

Figure 2.

Urinary VEGF levels measured at baseline, end of RT and one month post RT: Urine VEGF levels at baseline, end of radiation (RT) and 1 month (m) post-RT. Patients with undetectable VEGF levels were assumed to have a VEGF value of 24 pg/mg, the lowest detectable level. The ▪ represents the median value and the error bars represent the 95% confidence interval.

Only 10 patients had paired baseline and end of radiation samples. For analysis purposes, patients with an undetectable VEGF level were assumed to have a value of 24 pg/mg, the lowest detectable level of VEGF for our assay. 40% (4/10) of patients had a > 3-fold increase (relative to baseline) in urine VEGF. This is similar to the > 3-fold increase in plasma VEGF seen in 44% of metastatic renal cell cancer patients treated with sunitinib⁴².

Paired samples between baseline and 1 month post-radiation were available for 4 patients. We were unable to correlate change in urinary VEGF levels and radiation response because of the lack of paired samples. Of the paired samples, 2 of these patients had metastatic melanoma and continued on sunitinib following radiation. One had a > 2 fold increase in VEGF from baseline to 1 month post-radiation and developed progressive disease while the other had a 72% reduction in VEGF between these same time points and had stable disease. We also analyzed changes in VEGF levels between baseline and the end of radiation in patients who did and did not develop grade 3 hematologic toxicities. 3/5 patients with grade 3 hematologic toxicities had paired urine samples between baseline and the end of radiation while 5/5 patients without grade 3 hematologic toxicity had paired samples. The median VEGF fold change (end of radiation/baseline) between those with a grade 3 hematologic toxicity versus those without was 5.0 versus 0.75, (p=0.036) see Figure 3. The median radiation dose delivered to these two groups was not significantly different (p=0.39).

Figure 3.

Urinary VEGF fold change in patients with and without grade 3 hematologic toxicity: Urinary VEGF fold change (pg/mg) between baseline and end of radiation for patients who did not (n=5) and did (n=3) develop grade 3 or 4 hematologic toxicity during treatment. Median values for those who did not develop toxicity were significantly different from those who did develop toxicity (0.8 pg/mg range [0.4,2.2] vs. 5.0 pg/mg range [4.1, 17.8], p = 0.037, Wilcoxon Rank Sum exact test). Wilcoxon scores are plotted, which are based on the ranks of the original data.

Urine MMP-9

Urine MMP-9 was detectable in 21/27 (78%) samples. All 6 undetectable samples were from patients with metastatic melanoma and were over various time points (3 baseline, 2 end of RT, 1 one-month post-RT). The median MMP-9 concentration over all three time points for those with detectable levels was 2,095 pg/mg (range 80–804,101 pg/mg) and there was no significant difference between time points. To determine the median fold change for patients with paired samples from baseline and the end of RT samples that were undetectable were assumed to be 72 pg/mg, which is the lowest detectable level of MMP-9 for our assay. There was an increase in MMP-9 from baseline to the end of RT seen in 6/10 patients with a median fold increase of 21 (range 2–227 pg/mg).

Urine MET

Urine MET was detectable in 27/27 (100%) samples. Median MET concentration overall all three time points was 324 pg/mg (range 44–1109 pg/mg) with no significant difference between time points. There was an increase in MET from baseline to the end of RT seen in 5/10 patients with a median fold increase of 1.8 (range 1.7–21 pg/mg).

Urine MMP-2

Urine MMP-2 was detectable in 11/27 (41%) samples. The median MMP-2 concentration over all three time points for those with detectable levels was 878 pg/mg (range 199–10,359 pg/mg). Given the high number of samples with undetectable levels further analysis was not carried out on this biomarker.

Conclusions

This trial sought to assess the safety and tolerability of combining RT to intracranial tumors with continuously dosed sunitinib. Overall, the combination of sunitinib and RT was well tolerated. The majority of the toxicities were not severe, nor did they limit the patient’s ability to complete the combined course. Toxicity was limited to grade 1–2 in 40% of the patients in this study. Forty-seven percent of the patients in this study had grade 3 toxicities, most commonly hematologic and without significant clinical consequences. We were particularly interested in differentiating toxicity that occurred as a result of sunitinib, RT, or the combination to assess the potential for a more than additive toxicity profile. Among the 14 grade 3+ toxicities, 7 hematologic and metabolic toxicities were attributed to sunitinib, 2 were attributed to RT, 2 were attributed to the combination, 1 was attributed to disease progression and 2 venous thromboembolic events occurred which may have been related to sunitinib or to the underlying procoagulant status of patients with brain tumors. Grade 3 toxicity rate did not increase with increasing RT dose.

Comparing the experience from this trial with the known toxicity data of WBRT is possible in a limited manner. Over the past decade 4 prominent phase III trials have used a WBRT alone control arm: Radiation Therapy Oncology Group (RTOG) 0118⁴³, RSR-13⁴⁴, RTOG 9508⁴⁵, and motexafin gadolinium⁴⁶. When compared to these historic controls, the rate of treatment-related adverse events of continuously dosed sunitinib and intracranial RT may be modestly higher than the toxicity of WBRT alone, see Table 3.

There is only one other study that has evaluated the toxicity of continuously dosed sunitinib and radiation therapy: Kao et al.⁴⁷ recently published an experience treating 21 patients to 36 non-CNS oligometastatic sites with the combination of sunitinib and hypofractionated, image-guided RT (IGRT). They reported a 57% response rate and 28% rate of stable disease. They described a DLT in 1 of 10 patients when sunitinib 37.5 mg was combined with 10 Gy × 5 fractions and 2 of 5 patients had a DLT with sunitinib 50 mg combined with 10 Gy × 5 fractions. The 1-year local control, PFS, and OS rates were 85%, 44%, and 75%, respectively. They concluded that the combination of sunitinib (25–37.5 mg) to IGRT is tolerable in patients with oligometastatic disease without potentiation of RT-related toxicity. The rate of grade 3–5 toxicities of continuously dosed sunitinib and intracranial RT is similar to the rates seen in other larger studies that use sunitinib 37.5 mg continuous daily dosing alone. However, comparison is difficult because patients on continuously dosed sunitinib trials may have had a greater number of prior therapies and/or a longer duration of exposure to sunitinib.

A debate remains about the optimal timing of anti-angiogenic therapies and radiation therapy. Preclinical studies on inhibition of VEGF2 and the window of vascular normalization and radiosensitization by Winkler et al.⁴⁸ suggest an open window for maximum radiosensitivity in the range of 4–6 days after initiation of anti-VEGF2 therapies. In the clinical setting of glioblastoma multiforme, Batchelor et al.⁴⁹ described imaging techniques to detect vascular normalization after anti-angiogenic therapy. They discovered that vascular normalization may begin immediately upon anti-angiogenic therapy and has a window lasting approximately 28 days. The scheme for RTOG 0825, a phase III study testing radiation therapy and temozolomide with or without bevacizumab, starts bevacizumab after 30 Gy of the planned 60 Gy of radiation therapy has been delivered. Future clinic trials in this arena should employ advanced imaging techniques like diffusion-weighted MRI or permeability mapping through dynamic contrast-enhanced MRI to determine that optimal timing of concurrent anti-angiogenic therapy and radiation therapy. Alternately, the development of blood or urine-based biomarkers of vascular normalization could be developed to guide optimal timing of these therapies.

Concern about using antiangiogenic agents in patients with brain metastases has been raised by Pouessel and Culine,⁵⁰ however no patients from the present study experienced intracranial hemorrhage. Similarly, there was no intracranial hemorrhage in 15 patients with recurrent malignant glioma treated with bevacizumab and RT in a pilot study conducted at Memorial Sloan Kettering⁵¹.

When antiangiogenic drugs alter vascular permeability, they may change the apparent size of tumors without affecting the underlying tumor mass compromising our ability to visualize these tumors on MRI. Therefore, response rates as an indication of treatment efficacy should be cautiously interpreted when antiangiogenic agents are used.

Our patients had a median urine VEGF level, over all time points, of 336 pg/mg. This is similar to the mean urine VEGF level of 317 pg/mg seen at baseline for a cohort of patients with prostate, breast, brain, and hematologic malignancies³⁹. However other studies have reported median urine VEGF levels of 753 pg/ml at presentation for a cohort of pediatric and adult brain tumor patients and 101 pg/ml at baseline in patients with advanced soft tissue sarcomas^{38, 52}. These results emphasize the wide range of baseline urine VEGF levels seen for different histologies and stages of disease. It also suggests that looking at trends between different time points is likely to be more fruitful as a predictive marker than absolute value cut offs. However, given the limited number of patients with complete urine sets, we were unable to investigate which comparisons between time points might be most predictive of response.

We found a correlation between grade 3 hematologic toxicity and higher VEGF fold changes between baseline and the end of radiation. Significant changes in white blood cell counts during sunitinib treatment have been reported in patients with GIST⁵³. The largest proportional decrease was seen in monocytes which may reflect the fact that they have higher expression of VEGF receptor 1 relative to other peripheral blood mononuclear cells (PBMC)⁵³. Sunitinib related PBMC changes may also be related to “off-target” inhibition of KIT and FLT-3. Given the small data set, this finding is hypothesis generating and merits evaluation in future studies.

We are encouraged in this limited study by the responses seen. The 58% 6-month PFS of the patients with brain metastases is higher than the historic 6-month PFS of patients with brain metastases^54–56. The RTOG reported that brain metastases were the cause of death in 33%–50% of patients that were enrolled on studies 7916, 8528, and 8905⁵⁴. The 13% rate of neurologic death in this study also compares favorably to historic data. We are aware that patient selection can bias small studies, and this improvement will have to be validated by a larger study.

In conclusion, 37.5 mg daily dosing of sunitinib combined with cranial RT yielded acceptable toxicities and adverse events. This unique combination also produced intriguing responses. A phase II study further evaluating the use of sunitinib as a radiosensitizing agent in the setting of brain metastases is being planned by the RTOG.

Table 5.

Comparison of toxicities of trials utilizing whole brain RT compared with toxicity of RT combined with continuously dosed 37.5 mg sunitinib

	WBRT + Sunitinib	RTOG 950838	RTOG 011836	Motexafin Gadolinium39	Efaproxiral37
n	9	167	93	208	250
Grade 1–2 acute tox	40%	62%	66%	NR	NR
Grade 3 acute tox	47%	0%	27%	NR	NR
Grade 4 acute tox	0%	0%	5%	NR	11%
Grade 5 acute tox	13%	0%	2%	NR	NR
All grade nausea	27%	NR	NR	24%	NR
All grade diarrhea	13%	NR	NR	2%	NR
All grade paresthesia	13%	NR	NR	11%	NR
All grade pain	27%	NR	NR	8%	NR
All grade rash	13%	NR	NR	14%	NR
All grade tachycardia	0%	NR	NR	2%	NR
All grade peripheral edema	0%	NR	NR	11%	NR
Hematologic grade 3+	20%	NR	NR	NR	NR
Neurologic-grade 3+	20%	NR	12%	NR	3%
GI- grade 3+	0%	NR	6%	NR	6%
Nausea grade 3+	0%	NR	NR	NR	2%
Vomiting grade 3+	0%	NR	NR	NR	2%
Constipation grade 3+	0%	NR	NR	NR	2%
Vascular- grade 3+	13%	NR	4%	NR	NR
Constitutional-grade 3+	13%	NR	5%	NR	NR
Metabolic- grade 3+	13%	NR	3%	NR	NR