Bevacizumab (BVZ)-associated toxicities in children with recurrent central nervous system tumors treated with BVZ and irinotecan (CPT-11)

A Pediatric Brain Tumor Consortium Study (PBTC-022)




The incidence and spectrum of acute toxicities related to the use of bevacizumab (BVZ)-containing regimens in children are largely unknown. This report describes the adverse events in a recently completed large phase 2 trial of BVZ plus irinotecan (CPT-11) in children with recurrent central nervous system tumors.


Pediatric Brain Tumor Consortium trial-022 evaluated the efficacy and toxicity of BVZ (10 mg/kg administered intravenously) as a single agent for 2 doses given 2 weeks apart and then combined with CPT-11 every 2 weeks (1 course = 4 weeks) in children with recurrent central nervous system tumors. Children were treated until they experienced progressive disease, unacceptable toxicity or completed up to a maximum of 2 years of therapy. Toxicities were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 3.0. Patients who received at least 1 dose of BVZ were included for toxicity assessment.


Between October 2006 and June 2010, 92 patients evaluable for toxicity were enrolled and received 687 treatment courses. The most common toxicities attributable to BVZ included grade I-III hypertension (38% of patients), grade I-III fatigue (30%), grade I-II epistaxis (24%), and grade I-IV proteinuria (22%). Twenty-two patients (24%) stopped therapy due to toxicity.


The combination of BVZ and CPT-11 was fairly well-tolerated, and most severe BVZ-related toxicities were rare, self-limiting, and manageable. Cancer 2013;119:4180–4187. ©2013 American Cancer Society.


Although there have been many advances in the treatment of pediatric central nervous system (CNS) tumors, there remains a significant proportion of patients with refractory, progressive, or recurrent disease for which there are no known curative options. In fact, pediatric CNS tumors still account for the highest morbidity and mortality among pediatric malignancies.[1, 2] In an effort to improve outcomes, a variety of novel targeted therapies have been tested, including antiangiogenic agents.

Angiogenesis is defined simply as the development of new vasculature, and it is understood to be critically important for several physiologic and pathologic states including tumor growth, invasion, and metastases.[3-8] Vascular endothelial growth factor (VEGF) and its isoforms are perhaps the most important mediators of tumor angiogenesis, are overexpressed in a variety of human cancers, and may be associated with tumor progression.[9-12] In addition to its role in tumorigenesis, VEGF regulates many normal physiologic processes. It is crucial to angiogenesis during embryogenesis and throughout childhood and adulthood, and it is a regulator of vascular permeability.[13, 14] VEGF plays an important role in maintenance of blood pressure, wound healing, osteogenic differentiation, and maintenance of cells within the immune and nervous system.[14, 15] VEGF is also involved in glomerular endothelial integrity and renal protein handling under normal and abnormal physiologic conditions.[16] Finally, there are preliminary data to suggest that VEGF may be important in helping to regulate the coagulation cascade.[17] These normal physiologic processes regulated by VEGF are important in understanding many of the toxicities observed with anti-VEGF or VEGF-receptor therapies currently in clinical use.

Blocking tumor angiogenesis has become an increasingly attractive modality for anticancer therapy. Although there are many agents currently in use, bevacizumab (BVZ) (Avastin; Genentech Corp, San Francisco, Calif) has been widely studied in both pediatric and adult patients since its development in 1997.[5, 7, 18-24] BVZ is a humanized monoclonal immunoglobulin G1 antibody against all human VEGF isoforms and their proteolytic fragments.[25, 26] VEGF inhibition due to BVZ results in decreased vascular permeability, changes in interstitial fluid pressure, more orderly blood flow due to pruning of unnecessary blood vessels, decrease in the number of tumor-initiating cells due to disruption of their perivascular niche, and an increase in chemotherapy delivery into the tumor.[27, 28] However, the concomitant inhibition of endogenous circulating VEGF results in a predictable pattern of toxicities that has been well characterized in adults with recurrent solid tumors treated with BVZ or BVZ-containing regimens. The most common BVZ-associated toxicities in adults have included fatigue, epistaxis, hypertension, proteinuria, hemorrhage, poor wound healing, and thromboembolic events.[1, 29] Although children often have unique patterns of toxicities as compared to their adult counterparts, BVZ-related toxicities in smaller pediatric studies have been similar to what has been reported in the adult literature.[20, 30-33] Here, we report the observed toxicities in the largest cohort of evaluable children, to our knowledge, treated on a phase 2 trial of BVZ plus irinotecan (CPT-11) for recurrent CNS tumors.


The Pediatric Brain Tumor Consortium trial 022 (PBTC-022) was a phase 2 trial in children with recurrent CNS tumors using a combination of BVZ and CPT-11, in order to evaluate the efficacy and toxicity of this regimen. There were 5 separate strata comprising malignant glioma (stratum A), diffuse intrinsic pontine glioma (stratum B), medulloblastoma (stratum C), ependymoma (stratum D), and low-grade glioma (stratum E). Efficacy data for patients enrolled on strata A, B, D, and E have been previously reported.[34-36] Stratum C was prematurely closed due to slow accrual. All patients initially received 2 doses of BVZ at 10 mg/kg intravenously given 2 weeks apart. The first dose of CPT-11 was given within 3 days after the second dose of BVZ. For patients not on enzyme-inducing anticonvulsant drugs, the starting dose of CPT-11 was 125 mg/m2. This was increased in subsequent courses to150 mg/m2 if tolerated. For patients on enzyme-inducing anticonvulsant drugs, the starting dose was 250 mg/m2, which was increased by 25 mg/m2 increments every 2 weeks to a maximum of 350 mg/m2 as tolerated. The dose of CPT-11 was adjusted as previously reported based on hematologic and nonhematologic toxicities in subsequent courses.[34, 37] According to the protocol, patients who could not receive CPT-11 due to toxicity were able to continue treatment with BVZ alone in the absence of severe thrombocytopenia. BVZ was held for any grade III or higher toxicity at least possibly related to its use, and it could be restarted when toxicity had resolved to less than grade II. CPT-11 was not given if BVZ was held for toxicity. In general, patients were required to come off therapy if BVZ was held for toxicity and could not be restarted within 4 weeks. Toxicities were graded according to the National Cancer Institute (NCI) CTCAE, version 3.0. There were no dose reductions allowed for toxicities related to BVZ. Any patient who received at least 1 dose of BVZ was included for toxicity assessment.

Because of some of the specific known toxicities associated with BVZ use, there were routine surveillance laboratory testss and imaging required prior to and at designated time points throughout therapy. Because proteinuria is a known toxicity seen in adults and small pediatric series, it was required that urine protein be monitored by urine analysis for urine protein:creatinine ratio both prior to initiation of protocol therapy and every 4 weeks while on treatment. If the urine protein:creatinine ratio was greater than 1, protein excretion was quantified in a 24-hour urine collection and BVZ held if urine protein was greater than 3.5 grams (greater than grade III) and not restarted until it decreased to less than a grade III. If BVZ was held for greater than 4 weeks due to proteinuria, the patient came off study therapy. BVZ was discontinued and patients taken off study treatment if they developed nephrotic syndrome (grade IV proteinuria).

Also, due to the possible deleterious effects of epiphyseal growth resulting from VEGF inhibition by BVZ, patients were required to have a baseline radiograph of the left knee if they had not yet achieved their full growth potential.[38] This was repeated every 12 weeks thereafter during therapy or sooner if they had new symptoms of musculoskeletal pain. If any abnormalities were detected on routine radiographs following treatment, MRI scan of both knees and/or the affected area were performed.

Finally, given that the risk and the potential morbidity and mortality associated with a CNS hemorrhage, any evidence of a CNS hemorrhage (except very small punctate lesions) required that patients be taken off protocol therapy. Routine susceptibility-weighted and gradient-echo MRI sequences were required both prior to initiation of therapy and with each routine surveillance MRI scan throughout treatment.


Between October 2006 and June 2010, 97 patients were enrolled on the PBTC-022 study. Two patients were ineligible; 1 in stratum B and another in stratum D. Three other patients were eligible but not evaluable due to withdrawal before beginning protocol treatment (Table 1). Therefore, 92 patients were evaluable for toxicity and received a median of 4 courses (range, 1-26; 687 total treatment courses). The median age of all patients was 10.3 years (range, 0.6-20.1 years). The distribution of patients among the 5 strata is shown in Table 1. The most common BVZ-attributed toxicities included hemorrhage (mostly epistaxis), hypertension, fatigue, headache, and proteinuria (Table 2).

Table 1. Patient Accrual by Strata
StratumDiagnosisNo. of Patients EnrolledNo. of Patients Evaluable for ToxicityTotal Courses
  1. a

    Three patients (1 in stratum A and 2 in stratum E) were eligible but not evaluable due to withdrawal prior to beginning protocol treatment.

AMalignant glioma18a1783
BDiffuse intrinsic pontine glioma17 (1 ineligible)1653
DEpendymoma15 (1 ineligible)1458
ELow-grade glioma37a35433
Total 97 (2 ineligible)92687
Table 2. Most Commonly Reported Toxicities by Gradea
Toxicity (All Patients)Grade 1bGrade 2 bGrade 3 bGrade 4 bTotal EventsTotal Patients
  1. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; SGOT, serum glutamic oxaloacetic transaminase; SGPT, serum pyruvic oxaloacetic transaminase.

  2. a

    Some patients experienced more than 1 grade or episode of a particular toxicity which explains why there are sometimes more episodes and grades of a specific toxicity as compared to the total number of patients.

  3. b

    No. of events (no. of patients).

Emesis162 (41)84 (31)20 (12)0 (0)26651
Nausea97 (48)21 (14)2 (2)0 (0)12050
Diarrhea78 (38)12 (10)5 (5)0 (0)9539
Leukopenia63 (39)17 (11)2 (2)0 (0)8243
Lymphopenia39 (26)16 (13)16 (11)4 (3)7537
Hemorrhage71 (33)2 (2)1 (1)0 (0)7435
Hypertension47 (30)21 (20)3 (3)0 (0)7135
Fatigue50 (25)12 (8)1 (1)0 (0)6328
Proteinuria33 (20)24 (9)4 (3)1 (1)6220
Neutropenia15 (9)24 (18)15 (12)1 (1)5527
Pain (abdomen)47 (22)4 (3)1 (1)0 (0)5223
AST/SGOT36 (23)2 (2)2 (2)0 (0)4023
Anemia (hemoglobin)29 (19)10 (8)1 (1)0 (0)4021
ALT/SGPT27 (22)5 (3)1 (1)0 (0)3324
Thrombocytopenia24 (11)4 (3)0 (0)0 (0)2813
Pain (headache)17 (12)7 (6)4 (3)0 (0)2817
Hypocalcemia18 (12)3 (3)2 (2)0 (0)2315
Hypophosphatemia19 (11)2 (2)0 (0)0 (0)2119
Hypokalemia18 (15)0 (0)0 (0)0 (0)1815
Cerebrovascular ischemia0 (0)1 (1)0 (0)2 (2)33
Osteonecrosis1 (1)2 (2)0 (0)0 (0)33
Thrombo-embolism0 (0)1 (1)0 (0)0 (0)11
Poor wound healing0 (0)0 (0)0 (0)0 (0)00

Hemorrhage occurred in 35 patients (38%) (Table 3); 22 patients (24%) had grade I or II epistaxis that was self-limiting. There were also 10 patients (10.8%) with grade I CNS hemorrhage, 2 patients (2.1%) with grade I rectal bleeding, and 1 patient (1.1%) with grade III subdural hematoma (Table 3). Seventy-one episodes (mostly grade I-II) of systemic hypertension occurred in 35 (38%) patients (Table 2), of whom 3 also had episodes of grade III hypertension. Treatment of hypertension was at the discretion of the treating physician, but patients demonstrated good control of blood pressure with antihypertensive treatment, and no patient was taken off study for this toxicity alone. One patient, however, came off therapy due to a combination of hypertension and grade III/IV proteinuria. There were 63 episodes of mild to moderate fatigue among 28 (30.4%) patients with only one that worsened to grade III and required coming off therapy (Table 2; Table 4). Headache was reported in 17 patients (18.5%). One patient with recurrent ependymoma (stratum D) was taken off treatment due to persistent grade III headache following 11 courses of BVZ. There were 62 episodes of proteinuria among 20 patients (21.7%) with 4 episodes (among 3 patients) of grade III or IV proteinuria (Table 2). All patients with grade III/IV proteinuria were taken off therapy. Urine protein normalized over a period of 2 to 3 months following cessation of BVZ therapy in most patients. Three patients suffered cerebrovascular ischemia (1 grade II and 2 grades IV) while on treatment. One patient was asymptomatic with only surveillance neuroimaging findings of ischemia. Two other patients presented with neurologic deficits including hemiparesis, and ischemia was later confirmed on neuroimaging. All 3 of these children were taken off protocol therapy.

Table 3. All Hemorrhages by Location and Toxicity Gradeab
Hemorrhage TypeGrade 1Grade 2Grade 3Grade 4Total EventsTotal Patients
  1. a

    Some patients experienced more than 1 grade or episode of a particular toxicity which explains why there are sometimes more episodes and grades of a specific toxicity as compared to the total number of patients.

  2. b

    No. of events (no. of patients).

Pulmonary/upper respiratory/nose58 (21)2 (2)0 (0)0 (0)6022
Central nervous system10 (10)0 (0)0 (0)0 (0)1010
Rectum/gastrointestinal3 (2)0 (0)0 (0)0 (0)32
Subdural hematoma0 (0)0 (0)1 (1)0 (0)11

There were 3 episodes (1 grade I and 2 grade II) of osteonecrosis among 92 patients (3.3%).[39] These events occurred in the lunate bone of the wrist (1 patient) and in the femoral heads (2 patients), which are unique locations as compared with the adult literature, where most osteonecrosis due to BVZ occurs in the jaw bones.[39-42] All 3 children came off therapy.[39] One patient (1.1%) suffered a grade II vascular access-associated thromboembolism. Although no patients had delayed wound healing while on treatment, 2 patients with recurrent low-grade glioma who required a laminectomy and a shunt revision for hydrocephalus, respectively, were taken off therapy for concerns regarding the potential risk of delayed wound healing and were classified as off treatment due to “other complicating diseases” (Table 4). Another patient came off therapy at the “physician's discretion” due to skin ulcers and potential for poor wound healing. Twenty-five (27%) patients came off treatment due to toxicity or physician discretion. This included 21 patients (23%) who came off treatment due to BVZ-related toxicity, 3 patients who came off due to “other complicating diseases/physician discretion,” and 1 patient who came off due to pancreatitis attributed to CPT-11 (Table 4). BVZ-specific toxicities were evaluated over time in an exploratory fashion to assess if duration of BVZ exposure had an impact on the onset or severity of any toxicity. Only the incidence of proteinuria suggested a possible association with duration of BVZ exposure (Table 5).

Table 4. Toxicities/Events Requiring Patients to Come Off Study
StratumAdverse Event/Complication Requiring Patient to Come Off Study Treatment
  1. Abbreviation: CNS, central nervous system.

  2. a

    These were not observed toxic events but rather physician discretion or concern regarding a complicating concurrent disease.

ACNS ischemia
APhysician discretion: skin ulcersa
APancreatitis (attributed to CPT-11)
BCNS hemorrhage
BCNS hemorrhage
BCNS hemorrhage
BCNS ischemia
CCNS hemorrhage
ECNS ischemia
EComplicating diseasea
EProteinuria + hypertension
ERecurrent hip pain
EComplicating diseasea
ECNS hemorrhage
ECNS hemorrhage
Table 5. Frequency of Toxicities Based On Time of Exposure to Bevacizumab
Course No.No. of Patients Starting CourseNo. With HemorrhageNo. With HypertensionNo. With FatigueNo. With Proteinuria

Toxicities that occurred in this study that were probably related/attributable to CPT-11 included emesis (55.4%; n = 51, including 12 patients with grade III or IV), leukopenia (46.7%; n = 43, including 2 patients with grade III), neutropenia (43%; n = 40, including 13 patients with grade III or IV), diarrhea (42.3%; n = 39, including 5 patients with grade III or IV), and lymphopenia (40.2%; n = 37, including 14 patients with grade III or IV) (Table 2). As mentioned above, there was also one incident of pancreatitis that was attributed to CPT-11, requiring the patient come off therapy. There were no toxicity-related deaths.


Bevacizumab is being increasingly used in both adult and pediatric patients with cancer, often in the setting of recurrent disease and in combination with other chemotherapeutic and biological agents. The early reports of its toxicity profile in children were described among a small group of patients, and it is unclear if they depicted the true range and frequencies of BVZ toxicities.[20, 31, 33] Here, we present the largest prospective cohort of children with recurrent brain tumors treated with BVZ in combination with CPT-11.

The pattern of toxicities in this study is very similar to that observed in adult patients treated with BVZ.[1] Even the rarer BVZ-associated toxicities seen in our pediatric cohort such as cerebrovascular ischemia (3.3%), osteonecrosis (3.3%), and thrombosis (1.1%) have been described in adult patients.[15, 41-45] It is possible that there were other contributing factors that led to these adverse events in our cohort. For example, 2 of the 3 patients who developed cerebrovascular ischemia had previous radiation exposure which is a known risk-factor for vascular occlusion and strokes in childhood cancer survivors.[46] In a phase 2 study of BVZ given concurrently with radiotherapy and temozolomide in adults with newly diagnosed malignant glioma, a higher risk of CNS ischemia was observed with a pattern suggestive of involvement of small vessels, including lenticulostriate perforating arteries and potentiation of radiation-induced occlusive arteriopathy.[47] Similarly, exposure to steroids in 2 of 3 patients with osteonecrosis and the presence of a venous access device in the one patient with a subclavian vein thrombosis are known predisposing factors for these adverse events, even in patients not receiving BVZ. The radiographic findings of osteonecrosis do not differ when induced by steroids versus BVZ, so distinguishing which agent caused or contributed more to the abnormality is not possible. More than a third of the patients with low-grade gliomas who were enrolled on stratum E had longer exposure to BVZ than did patients on the pediatric phase 1 study.[48] It is possible that this prolonged exposure to BVZ may have led to more frequent toxicities in our population.[48]

Fatigue occurred in approximately 30% of patients, but only one patient came off of study treatment due to this toxicity. Interestingly, a recent report showed that 7 of 30 children developed new-onset hypothyroidism requiring thyroid hormone replacement after initiating BVZ, although many of these patients had previously received cranial radiation as well.[33] Unfortunately, this toxicity was not well established during the time period our patients were accrued and not routinely tested for, so the extent that undiagnosed hypothyroidism contributed to our patients' fatigue is unknown.

Most CNS hemorrhages, although commonly observed, were only mild and asymptomatic. These small CNS bleeds may have been recognized more readily due to the frequency of imaging and use of routine susceptibility-weighted and gradient-echo MRI sequences that demonstrate smaller hemorrhages more readily.[49] In the patient with the grade 3 subdural hematoma, the event occurred approximately 3 weeks after the patient underwent a surgical procedure for the insertion of a vagal nerve stimulator to control seizures. The patient's scheduled BVZ was held prior to this procedure; however, both the BVZ and the procedure itself may have added to the patient's risk for bleeding. The risk of hemorrhage (all locations) is increased in adult patients receiving BVZ. In a meta-analysis, BVZ significantly increased the risk of bleeding when compared with controls, and there was a significantly increased risk for epistaxis or pulmonary hemorrhage in patients with cancer who receive BVZ.[50] The risk of a fatal hemorrhage was low, but patients with CNS tumors were not represented in the analysis.[50] The risk of CNS hemorrhage is of major concern for pediatric brain tumor patients who receive BVZ. Although bleeding occurred in 38% of our patients, the vast majority of these hemorrhages consisted of self-resolving grade I and II epistaxis. It is unclear if there is a significant increase in intracranial hemorrhages in patients with CNS tumors who receive BVZ.[50] In a randomized study evaluating the efficacy of BVZ alone versus BVZ plus CPT-11 in adults with recurrent glioblastoma multiforme, intracranial hemorrhage was seen in only 2 of 85 (2.4%) patients receiving BVZ alone compared with 3 of 82 (3.8%) patients receiving BVZ plus CPT-11.[19] Although it is theoretically possible that BVZ could increase the risk of CNS bleeding in patients with brain tumors, it is unclear if the occurrence of intracranial hemorrhage is significantly higher than the inherent risk of this complication in the CNS tumor population, especially in the context of prior radiotherapy and/or progressive disease. Still, serious intracranial hemorrhages must be considered a potential adverse event in both adults and children who receive BVZ therapy.

Hypertension and proteinuria were no more common in children than in adults, and hypertension readily responded to antihypertensive interventions in most patients. Both proteinuria and hypertension have been reported as dose-dependent within the adult literature and seen among 41% to 63% of adult patients treated with BVZ.[51, 52] Although the exact mechanism of these renal toxicities is not fully understood, thrombotic microangiopathy has been described in some adult patients who have undergone renal biopsies.[53] Despite the high incidence of proteinuria, most cases are asymptomatic and not severe, with nephrotic range proteinuria reported in less than 10% of adult patients.[52] Although rare in our series as well, grade III or IV proteinuria did occur, and all grades of proteinuria seemingly occurred more often in patients who received BVZ for a prolonged duration (Table 5). Because of these risks, all patients receiving BVZ should be monitored carefully with quantitative urine protein measurements at least monthly during therapy in addition to routine blood pressure assessments.

BVZ is increasingly being used in pediatric studies to treat children with CNS tumors and non-CNS tumors alike, both in the upfront and recurrent setting. A better understanding of its toxicity profile is essential to not only inform patients and their families of the risks but also to detect and treat adverse events in a timely fashion, thereby preventing late sequelae. Our experience within the confines of this study shows that BVZ in combination with CPT-11 was fairly well-tolerated, and most adverse events were manageable and quite similar to those described in the adult population.


This trial was supported in part by NIH grant U01 CA81457 for the Pediatric Brain Tumor Consortium and by the American Lebanese Associated Charities.


Dr. Goldman served on an advisory board for Novartis. Dr. Friedman has received consulting fees/honoraria from Genentech. Dr. Gururangan has been a consultant for Boehringer Ingelheim. Dr. Banerjee has received grant money (salary support) and travel funds to PBTC meetings from the Pediatric Brain Tumor Consortium. All other authors made no disclosure.