Bevacizumab in high-grade glioma patients following intraparenchymal hemorrhage

Abstract

Background

Intraparenchymal hemorrhage (IPH) is a relative contraindication to bevacizumab therapy, an anti-vascular endothelial growth factor (VEGF) monoclonal antibody approved for the treatment of recurrent glioblastoma. However, in patients with symptomatic enhancing tumors and poor functional status, bevacizumab may be the only beneficial therapeutic option.

Methods

We retrospectively reviewed all patients with high-grade glioma who were treated between January 1, 2005 and December 31, 2014 with bevacizumab despite prior IPH.

Results

Eighteen patients met our study criteria. There were 12 women and 6 men with a median age of 56 years. Tumor types were glioblastoma (n = 15), anaplastic astrocytoma (n = 2), and anaplastic oligodendroglioma (n = 1). Seventeen patients had prior spontaneous intratumoral bleed (13 grade 1–2; 4 grade 3–4); the 1 remaining patient had a grade 3 bleed due to cerebral venous thrombosis. Among them, identifiable risk factors for hemorrhage were anti-VEGF therapy, anticoagulation use, thrombocytopenia, and hypertension; seven had no identifiable risk factors. The median duration from IPH to (re-)initiation of bevacizumab was 113 days (range 13–1367). Brain imaging performed prior to bevacizumab treatment showed persistent or evolving hemorrhage in 8 patients and complete resolution in 10 patients. With a median follow-up duration of 137 days after bevacizumab re-initiation, only 1 (6%) of the 18 patients re-bled; this patient had an anaplastic oligodendroglioma and developed a grade 2 intratumoral bleed after 3 doses of bevacizumab.

Conclusions

The incidence of re-bleed is rare. Bevacizumab use was safe in patients with recurrent high-grade glioma following IPH for whom no other meaningful treatment options existed.

Bevacizumab is a recombinant humanized monoclonal antibody that targets vascular endothelial growth factor (VEGF) and is approved for the treatment of recurrent glioblastoma (GBM) in the United States.^1–3 VEGF induces proliferation of vascular endothelial cells and is highly expressed in GBM.² Bevacizumab inhibits VEGF, thereby disrupting angiogenesis and tumor growth, and has been shown to increase progression-free survival and improve symptoms in patients with recurrent GBM.^3,4

Early clinical studies of bevacizumab demonstrated an increased risk of hemorrhage at both primary and metastatic cancer sites and bleeding-related mortality.^5–7 The mechanism underlying this bleeding risk is unclear, but is likely related to bevacizumab disrupting endothelial cell integrity in the tumor microvasculature.⁸ In light of these data, subsequent clinical trials excluded patients with a history of bleeding.^3,4,9,10 Consequently, a history of intraparenchymal hemorrhage (IPH) is widely considered a contraindication to bevacizumab therapy in patients with malignant brain tumors.

However, in such patients with symptomatic enhancing tumors and poor functional status, bevacizumab may be the only beneficial therapeutic option. The actual re-bleeding risk associated with bevacizumab is unknown, and it may be unjustified to deprive patients of bevacizumab therapy because of a prior IPH. The aim of this study was to determine the safety of treatment and incidence of recurrent IPH in patients with recurrent high-grade glioma who received bevacizumab therapy following prior IPH.

Materials and Methods

This was an IRB-approved, single-institution, retrospective review. We reviewed all patients with high-grade glioma who had IPH and were subsequently treated with bevacizumab between January 1, 2005 and December 31, 2014 at our institution. Patients who developed their first IPH after bevacizumab therapy were excluded unless they were re-challenged with bevacizumab. We reviewed the medical records of all patients who met inclusion criteria and available neuroimaging studies at 3 time points: 1) at initial IPH, 2) immediately before bevacizumab therapy, and 3) at last available imaging after (re-)treatment with bevacizumab or any post-bevacizumab imaging that revealed an IPH. We estimated the IPH volume using the ABC/2 method, where A is the greatest hemorrhage diameter, B is the diameter perpendicular to A, and C is the approximate number of imaging slices with hemorrhage multiplied by the slice thickness.¹¹ This method has been applied and validated for volumetric measurement of IPH and stroke using both computed tomography (CT) and magnetic resonance imaging (MRI) scans.^11–15 All IPHs were graded according to the common terminology criteria for adverse events (CTCAE) version 4.0. We stratified the duration from prior IPH to initiation of bevacizumab into acute (<30 days), subacute (30 to 90 days), and chronic (>90 days) phases.

Results

Patient Characteristics

We identified 18 patients who met our inclusion criteria (Table 1). There were 12 women and 6 men with a median age of 56 years. At the time of prior IPH, 15 patients had GBM, 2 had anaplastic astrocytoma, and 1 had anaplastic oligodendroglioma. At subsequent tumor recurrence, the 2 anaplastic astrocytomas progressed to GBM, while the anaplastic oligodendroglioma remained as an anaplastic oligodendroglioma. All 3 tumors were histologically confirmed.

Table 1.

Summary of Study Patients

Pt	Duration from prior IPH to Bev (days)	At Prior Intraparenchymal Hemorrhage (IPH)						At (Re-)initiation of Bev
		Tumor Histology	Tumor Treatment	Risk Factor for Bleed	IPH Etiology	IPH Volume (cm3)	IPH Grade	Recurrent Tumor Histology	Persistent IPH on pre-Bev scan	Re-bleed
1	13	GBM	Inv D	HTN	IT	1.78	4	GBM	Y	N
2	21	GBM	Inv D	–	IT	5.30	2	GBM	Y	N
3	22	GBM	Bev + TMZ	AC	IT	0.10	2	GBM	Y	N
4	43	GBM	Bev + TMZ	T grade 3	IT	2.07	2	GBM	N	N
5	56	GBM	Inv D (V)	AC	IT	0.29	2	GBM	Y	N
6	57	GBM	None	–	IT	0.36	2	GBM	N	N
7	61	GBM	RT + TMZ	AC	IT	0.40	2	GBM	Y	N
8	65	AA	TMZ	–	IT	7.06	3	GBM	N	N
9	86	GBM	TMZ	–	IT	0.10	3	GBM	N	N
10	140	GBM	Bev	AC	IT	50.19	2	GBM	Y	N
11	144	GBM	RT + TMZ	HTN	IT	2.32	3	GBM	Y	N
12	169	GBM	None	HTN	IT	0.08	3	GBM	N	N
13	201	AA	RT + TMZ	–	IT	2.51	3	GBM	N	N
14	211	GBM	RT + TMZ	T grade 2	IT	8.53	2	GBM	N	N
15	471	GBM	RT + TMZ	–	IT	NA	3	GBM	N	N
16	502	GBM	TMZ	AC + HTN	CVT	26.67	3	GBM	N	N
17	1005	GBM	RT	–	IT	0.04	1	GBM	N	N
18	1367	AO	TMZ	–	IT	34.92	3	AO	Y	Y

Abbreviations: AA, anaplastic astrocytoma; AC, anticoagulation; AO, anaplastic oligodendroglioma; Bev, bevacizumab; CVT, cerebral venous thrombosis; GBM, glioblastoma; HTN, hypertension; Inv D, investigational drug; Inv D (V), anti-vascular endothelial growth factor investigational drug; IT, intratumoral; NA, imaging not available for review; Plt, platelet level; Pt, patient; POD, progression of disease; RT, radiation therapy; T, thrombocytopenia; TMZ, temozolomide.

Prior IPH

Seventeen patients had spontaneous intratumoral bleed and 1 patient had IPH due to cerebral venous thrombosis. Of the 17 patients with intratumoral bleed, 9 had at least 1 additional identifiable risk factor for hemorrhage including treatment with an anti-VEGF drug (bevacizumab: n = 3; investigational anti-angiogenic drug: n = 1), abnormal coagulation due to concurrent use of anticoagulation for systemic venous thrombosis (enoxaparin: n = 4; dalteparin: n = 1) and/or presence of thrombocytopenia (n = 2), and presence of hypertension defined as systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg at the time of IPH (n = 4); the remaining 8 had no risk factors identified. At the time of prior IPH, 1 patient with a massive supratentorial GBM that extended into the brainstem developed dysphagia and stridor requiring urgent intubation for airway protection, meeting criteria for a grade 4 IPH (Fig. 1). Eight patients had grade 3 IPH and required an invasive intervention, including tumor resection in 4, surgical evacuation of the bleed in 3, and intraarterial thrombolysis in the 1 patient with cerebral venous thrombosis. Of the remaining 9 patients, 8 had a grade 2 IPH and were hospitalized for medical monitoring, and 1 with a grade 1 IPH was asymptomatic and was managed as an outpatient.

Fig. 1.

Noncontrast CT (A) and MRI axial SWAN (B) images demonstrating a left temporal intraparenchymal hemorrhage in a patient with recurrent glioblastoma who required intubation for airway protection. Axial FLAIR (C) and contrast T1-weighted (D) images show extensive tumor involving bilateral hemispheres and brainstem associated with contrast enhancement. This patient was extubated successfully following bevacizumab therapy. Abbreviations: SWAN, susceptibility-weighted angiography; FLAIR, fluid-attenuated inversion recovery.

Bevacizumab Treatment

Three patients were treated with bevacizumab during the acute phase of prior IPH (<30 days); including the 1 patient who was intubated. Six patients received bevacizumab in the subacute phase (30 to 90 days) and 9 received it in the chronic phase (>90 days). Of the 5 patients who were on anticoagulation during prior IPH, 3 resumed anticoagulation prior to bevacizumab therapy. The median duration from prior IPH to (re-)initiation of bevacizumab therapy was 113 days (range 13 to 1367). Brain imaging performed immediately prior to bevacizumab treatment showed persistent or resolving hemorrhage in 8 patients and complete resolution of prior IPH in 10.

Seventeen patients, including 3 patients who had prior IPH during anti-VEGF treatment, were treated with standard bevacizumab doses of 10 mg/kg every 2 weeks or 15 mg/kg every 3 weeks. One patient, who had developed a prior IPH while on bevacizumab, was re-challenged at an initial dose of 5 mg/kg, which was increased to 7.5 mg/kg on the second dose; she developed a wound infection and bevacizumab was discontinued after the 2 doses. In the whole cohort, a median of 4 doses (range 1 to 19) of bevacizumab was administered. All except 1 patient had symptomatic improvement after bevacizumab; the 1 patient had further disease progression.

Outcome Following Bevacizumab Treatment

With a median follow-up of 137 days after bevacizumab initiation, only 1 (6%) of the 18 patients developed a re-bleed (Fig. 2). This was the 1 patient with anaplastic oligodendroglioma who was started on bevacizumab 1367 days after prior IPH (chronic phase); she developed a grade 2 intratumoral bleed that was associated with further tumor progression after 3 doses of bevacizumab at 10 mg/kg every 2 weeks. This patient did not have any identifiable risk factors for bleed during prior IPH.

Fig. 2.

MRIs demonstrating prior and new IPH in the patient who developed a re-bleed following bevacizumab therapy. Prior IPH: Axial GRE (A), FLAIR (B) and contrast T1-weighted (C) images reveal hemorrhage within a heterogeneously enhancing tumor in the left fronto-parietal region. New IPH: Axial SWI (D), FLAIR (E) and contrast T1-weighted (F) images demonstrating a new focal intratumoral bleed in the left parietal region (arrow), away from prior IPH site. Abbreviations: IPH, intraparenchymal hemorrhage; GRE, gradiant echo; FLAIR, fluid-attenuated inversion recovery; SWI, susceptibility-weighted imaging.

Discussion

Prior IPH is considered a contraindication to the use of bevacizumab, which may prevent some patients from receiving this beneficial agent. Our study, to the best of our knowledge, is the first study to examine the risk of re-bleeding in high-grade glioma patients who were treated with bevacizumab despite prior IPH. Only 1 (6%) of 18 patients developed a re-bleed, and even this patient did not suffer any significant clinical compromise from the hemorrhage.

Spontaneous IPH has been associated with GBM and oligodendroglioma of all grades.¹⁶ In a study of 100 gliomas, 2 oligodendrogliomas and 3 GBMs (but no astrocytomas) developed clinically significant hemorrhage; of the remaining tumors, microhemorrhages were found in 56.7% of oligodendrogliomas, 53% of GBMs, and 10% of astrocytomas.¹⁷ Subsequent studies also showed a higher incidence of spontaneous intratumoral hemorrhage in oligodendrogliomas and GBM than in astrocytomas.¹⁸ Oligodendrogliomas have abundant vessel formation where the vascular structure is composed mainly of thin-wall capillaries; those with retiform vessels and calcified capillaries are associated with an increase risk of bleeding.¹⁷ In the 1 patient with re-bleed in our study, the presence of blood products 1367 days after the prior IPH suggested that the oligodendroglioma had continual intratumoral bleed because they otherwise would not have been present more than 3 years after the prior IPH. Seven other patients who had persistent blood products on their brain imaging all had GBMs and did not have re-bleeds. It is hence highly likely that the 1 re-bleed was due to the intrinsic propensity of the tumor to hemorrhage and less to the bevacizumab therapy.

Despite initial concerns of life-threatening hemorrhage with anti-VEGF use in patients with brain tumor, a review of 10 598 cancer patients in 57 clinical trials of anti-VEGF therapy, including bevacizumab, showed that the rate of intracranial hemorrhage, even in patients with high-grade glioma and brain metastases, was negligible (<1%).¹⁹ A subsequent review at Memorial Sloan Kettering Cancer Center showed an IPH frequency of 3.7% in cancer patients receiving bevacizumab, which was identical to a 3.6% frequency detected in comparable patients not treated with bevacizumab.²⁰ This finding was similar to that from a more recent study by the German Glioma Network, which demonstrated no significant difference in the rate of IPH with and without bevacizumab therapy (P = .571).²¹ In fact, one other study suggested that intracranial hemorrhage in high-grade glioma was due to tumor progression rather than anti-VEGF therapy.²² Overall, these studies support our view that the 1 case of re-bleed in our study was likely related to the tumor itself, and less likely due to bevacizumab therapy. In addition, all 4 patients who had prior bleed while on anti-VEGF therapy did not develop a re-bleed with bevacizumab therapy. Although the numbers are too small to draw any definitive conclusion, this provides preliminary evidence that re-challenge with bevacizumab therapy may be safe.

Our study has several limitations. First, this is a retrospective review with its inherent selection bias. Second, we estimated IPH volume with the ABC/2 method, which could potentially overestimate the bleed volume in patients on anticoagulation and when MRI was used.^13,23 However, this potential bias does not alter our findings because the single event of re-bleed does not allow any correlation analysis with IPH volumes.

Conclusion

In summary, bevacizumab treatment was safe in patients with recurrent high-grade glioma following IPH. Patients with prior IPH should be considered for bevacizumab therapy after proper counseling of re-bleeding risk, especially when no other meaningful treatment options exist. Future trials of bevacizumab in recurrent high-grade glioma patients should not exclude those with prior IPH.

Funding

This research was supported by a grant from the NIH (P30-CA008748).