A phase 2 study of bevacizumab with cisplatin plus intensity-modulated radiation therapy for stage III/IVB head and neck squamous cell cancer




For patients with stage III through IVB head and neck squamous cell carcinoma (HNSCC), concurrent high-dose cisplatin plus radiation therapy is a widely accepted standard of care. HNSCC tumors that express high levels of vascular endothelial growth factor have been associated with a worse prognosis, and bevacizumab may sensitize tumors to cisplatin and radiation.


Planned treatment consisted of definitive intensity-modulated radiation therapy (IMRT) (total, 70 grays) with concurrent cisplatin (50 mg/m2 on days 1, 2, 22, 23, 43, and 44) and bevacizumab (15 mg/kg on days 1, 22, and 43). The primary endpoint was 2-year progression-free survival (PFS), and overall survival (OS) was a secondary endpoint.


Forty-two previously untreated patients (34 men and 8 women; median age, 55 years; range, 27-75 years) with stage III through IV HNSCC without distant metastasis (oropharyngeal carcinoma, 39 patients; laryngeal carcinoma, 3 patients) were treated. Human papillomavirus (HPV) status by was determined by in situ hybridization (HPV positive, 16 patients; HPV negative, 14 patients, unknown HPV status, 12 patients). The toxicities (determined according to version 3.0 of Common Terminology Criteria for Adverse Events Common) that were experienced by all patients (any grade) were mucositis, lymphopenia, leukopenia, throat pain, fatigue, and anemia. There were 2 treatment-related deaths, including 1 sudden death and 1 death from aspiration pneumonia. The median follow-up was approximately 31.8 months (range, <3 to 51 months). The 2-year PFS rate was 75.9% (95% confidence interval, 63.9%-90.1%), and the 2-year OS rate was 88% (95% confidence interval, 78.6%-98.4%). Among 32 patients for whom post-treatment Head and Neck Performance Status Scores were obtained (median, 5.6 months after completing radiation therapy), scores of 100 for eating, speech, and diet, respectively, were recorded among 75%, 84%, and 50% of patients.


The addition of bevacizumab to high-dose cisplatin plus IMRT did not appear to increase toxicity to unacceptable levels among patients with HNSCC, and the efficacy results were encouraging. Cancer 2012. © 2012 American Cancer Society.


For patients with locally and/or regionally advanced head and neck squamous cell carcinoma (HNSCC), concurrent high-dose cisplatin plus radiation therapy is a widely accepted standard of care.1-5 The acute and chronic treatment-related toxicities of this regimen limit the addition of another cytotoxic agent, but the addition of targeted (or biologic) agents to concurrent chemoradiation is a strategy to potentially improve efficacy while maintaining tolerability.

Bevacizumab, a recombinant humanized monoclonal antibody that binds vascular endothelial growth factor A (VEGF-A), was the first antiangiogenesis agent to receive US Food and Drug Administration approval in oncology.6 At the time the randomized studies were ongoing that led to the approval of bevacizumab as part of treatment regimens for certain patients with advanced colon cancer and nonsmall cell lung cancer,6 preclinical and early clinical data accumulated in support of the study of this agent in patients with HNSCC. In a meta-analysis of 12 studies that evaluated VEGF-A expression in HNSCC (n = 1002 patients), tumors with positive VEGF status were associated with a 1.88-fold greater risk of death at 2 years.7 The inhibition of antiangiogenesis achieved radiosensitization in preclinical models of solid tumors,8-10 suggesting that clinical studies incorporating VEGF inhibition should be developed for cancers that are commonly treated with radiotherapy, such as HNSCC.

A phase 1 trial at the University of Chicago demonstrated that it is feasible to add bevacizumab (10 mg/kg every 2 weeks) to 5-fluorouracil (5-FU) plus hydroxyurea (FHX) with radiation therapy for patients with poor prognosis HNSCC.11 Most patients in this study had received previous radiation therapy for head and neck cancer. The recommended chemotherapy doses to be given with radiotherapy were oral hydroxyurea 500 mg twice daily, 5-FU 600 mg/m2 daily as a continuous infusion on days 1 through 5, and bevacizumab 10 mg/kg on day 1 of a 14-day cycle. These reductions in the doses of 5-FU and hydroxyurea, compared with the standard FHX regimen, were required because of increased neutropenia associated with the addition of bevacizumab. There were 2 fatal hemorrhages among 43 patients, which was similar to the incidence of hemorrhage described in previous reirradiation HNSCC studies.11 In contrast to an early report in which bevacizumab was associated with an increased risk of hemorrhage in squamous cell carcinomas of the lung,12 this phase 1 study provided reassurance that bevacizumab does not appear to be associated with an increased risk of hemorrhage in patients with HNSCC.11

Although various chemotherapeutic agents have been combined with radiation therapy for patients with newly diagnosed stage III through IVB HNSCC, cisplatin is the drug with the most data to support its use in this setting.1-4 The addition of bevacizumab to standard cisplatin-based chemoradiotherapy in this patient population was deemed an appropriate topic for clinical study, both because of the apparent tolerability of VEGF inhibition among HNSCC patients11, 13 and because of the radiosensitizing potential of VEGF inhibition.8-10 We designed and conducted a single-institution, nonrandomized, phase 2 study in which bevacizumab (15 mg/kg intravenously every 3 weeks) was added to high-dose cisplatin and concurrent radiation therapy (intensity-modulated radiation therapy [IMRT]).


This single-institution, nonrandomized, open-label, phase 2 study was approved by the Institutional Review Board of Memorial Sloan-Kettering Cancer Center.

Patient Eligibility and Baseline Assessment

Eligible patients had previously untreated, stage III through IVB HNSCC without distant metastasis (M0). Other key inclusion criteria were a Karnofsky performance status ≥70%, adequate bone marrow function (absolute neutrophil count, ≥1500/μL; platelets, ≥100,000/μL; hemoglobin, ≥9 g/dL), adequate renal function (serum creatinine ≤1.5 mg/dL and/or calculated creatinine clearance ≥55 mL per minute [calculated according to the Cockcroft and Gault equation]), and adequate hepatic function (total bilirubin ≤1.5 times the upper normal limit [UNL], aspartate aminotransferase ≤1.5 times the UNL, alanine aminotransferase ≤1.5 times the UNL, and alkaline phosphatase ≤1.5 times the UNL).

Key exclusion criteria were: hearing loss or multifocal peripheral sensory alterations or paresthesias interfering with activities of daily living; a history of arterial thromboembolic events within the last 3 years; a urine protein:creatinine ratio ≥1.0; current use of warfarin, heparin, or low-molecular weight heparin; chronic daily treatment with aspirin (>325 mg daily) or nonsteroidal anti-inflammatory medications known to inhibit platelet function; gross hemoptysis or hematemesis (defined as bright red blood ≥1 teaspoon) within 28 days before day 0 of protocol treatment; an anatomic lesion that increased the risk of serious hemorrhage, such as encasement or invasion of major blood vessels by primary tumor and/or by involved lymph nodes; blood pressure >150/100 mm Hg; major surgical procedure or significant traumatic injury within 28 days before treatment with bevacizumab; or a core biopsy within 15 days before treatment with bevacizumab.

All patients provided written informed consent before enrollment in the study. A percutaneous gastrostomy (PEG) tube was placed pretreatment for all patients, consistent with practice at our institution at the time of the conduct of this study. For patients who had adequate tissue available, the presence of high-risk human papilloma virus (HPV) strains (including HPV type 16) was determined by in situ hybridization using the Ventana HPV Family 16 probe (Ventana Medical Systems, Inc., Tucson, Ariz).14

Study Treatment

Chemotherapy treatment consisted of planned cisplatin (50 mg/m2 on days 1, 2, 22, 23, 43, and 44) and bevacizumab (15 mg/kg on days 1, 22, and 43). Each planned 21-day chemotherapy treatment period was defined as a cycle. Cisplatin and/or bevacizumab could be administered up to 3 days before the scheduled date if necessary for medical or personal reasons. The treatment schema is illustrated in Figure 1. The initial version of the protocol called for an additional 6 months of maintenance bevacizumab after chemoradiation, but this was discontinued in an amendment after a grade 4 pulmonary hemorrhage event in Patient 1 during maintenance treatment.

Figure 1.

The treatment schema is shown. HNSCC indicates head and neck squamous cell cancer; IMRT, intensity-modulated radiotherapy; FDG-PET, fluorodeoxyglucose-positron emission tomography; CT, computed tomography; MR, magnetic resonance.

Cisplatin dose reduction was allowed for grade ≥3 adverse events, and specific guidelines were provided for common cisplatin-related toxicities (hematologic, renal, auditory, and neurologic toxicities). Cisplatin dose level −1 was 50 mg/m2 only on the first day of the cycle, which represented a 50% dose reduction. Bevacizumab would be discontinued for any grade ≥2 hemorrhagic event or any thrombotic event requiring anticoagulation. Bevacizumab dose reductions were not allowed, but bevacizumab would be held for any toxicity grade ≥3 until resolution to grade ≤1. Dose modification in chemotherapy did not mandate any dose interruption or modification in planned radiotherapy.

Concurrent with chemotherapy, IMRT was delivered on a standard schedule of 5 days per week (excluding weekends and holidays) in 1 fraction daily of 212 centigrays (cGy) up to a total dose of 7000 cGy. Patients with high-risk subclinical disease received a total of 5940 cGy, whereas patients with lower risk subclinical disease received 5400 cGy. All treatments were given over approximately 33 days using an IMRT dose-painting technique.

Evaluation During Study and Response Assessment

Adverse events during treatment were assessed at weekly clinic visits using version 3.0 of the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE). Between 3 and 4 months after completion of chemoradiation, radiologic assessment was performed (whole-body fluorodeoxyglucose-positron emission tomography/computed tomography scan plus computed tomography or magnetic resonance imaging studies of the neck). Patients who had evidence of persistent disease in the head and neck after chemoradiation would have surgery planned for ≥60 days after the last bevacizumab infusion. For patients who maintained active follow-up after chemoradiation, clinic visits took place at least every 3 months for 2 years, every 6 months for the next 3 years, and annually thereafter.

Study Objectives and Statistical Considerations

The primary objective of this study was to determine 2-year progression-free survival (PFS) for patients with locally or regionally advanced HNSCC who received concurrent IMRT, cisplatin, and bevacizumab. The secondary objective was to determine the median overall survival (OS) of these patients. All patients who received any treatment on study were included in the efficacy and safety analyses.

At the time this study was designed, a review of concurrent chemoradiotherapy supported using a 45% rate as the baseline estimate for 2-year PFS.1, 3, 15-17 To detect an improvement in the 2-year PFS rate from 45% to 65%, the study required 42 patients (5% type I error; 20% type II error). Time endpoints were measured from the start of treatment until the time of first reported distant or local/regional failure or until death from any cause, consistent with published HNSCC chemoradiation literature.17 Data were censored at the time of last follow-up for patients without distant or local/regional failure. PFS and the median OS were estimated by using the Kaplan-Meier method.

The safety and tolerability of bevacizumab were central concerns during the design of the protocol. The study included an early stopping rule for any unanticipated toxicity, and criteria for closure of the study according to the early stopping rule were not met during the conduct of this study.


Patient Characteristics

Forty-four patients were enrolled on the study between November 20, 2006 and October 22, 2008. Two patients withdrew consent before receiving any treatment on the study, and the results detailed below pertain to the 42 patients who received any treatment on study.

Most patients were men (81%), the primary tumor site was oropharynx in 93% of patients, and most patients had stage IVA disease (76%). HPV DNA was detectable in the majority of tumors for which adequate material was available for in situ hybridization analysis. Twenty-nine patients (69%) had <10 pack-years of tobacco exposure. Table 1 lists baseline clinical data for the treatment population.

Table 1. Patient Characteristics
CharacteristicNo. of Patients (%)
  1. Abbreviations: HPV, human papillomavirus; KPS, Karnofsky performance status.

Age: Median [range], y55 [27-75]
 Men34 (81)
 Women8 (19)
KPS: Median [range]90 [70 to ≥90]
 Oropharynx39 (93)
  Base of tongue21
  Soft palate1
 Larynx3 (7)
 III10 (24)
 IVA32 (76)
HPV status 
 Positive19 (45)
 Negative14 (33)
 Unknown/insufficient material for assay9 (21)
Tobacco use, pack-years 
 0 to ≤121 (50)
 >1 to ≤108 (19)
 >10 to ≤206 (14)
 >207 (17)

Treatment Delivery and Acute Adverse Events

All patients received the full course of radiotherapy. Only 22 of 42 patients (52%) were able to receive the planned 3 cycles of bevacizumab plus cisplatin at the full dose because of toxicity. However, 37 of 42 patients (88%) were able to receive at least 2 of the 3 planned cycles of bevacizumab and cisplatin at the full dose.

Table 2 lists the adverse events that occurred in ≥30% of treated patients (regardless of grade) or events that were grade ≥3 in at least 2 patients and were possibly, probably, or definitely related to bevacizumab. Adverse events (any grade) experienced by all patients were mucositis (functional), lymphopenia, leukopenia, throat pain, fatigue, and low hemoglobin. The most common grade ≥3 adverse events were lymphopenia (100% of patients), mucositis (86% of patients), throat pain (74% of patients), and neutropenia (43% of patients). There were 2 deaths within 90 days of the last treatment, including 1 death from aspiration pneumonia and 1 sudden death of unclear etiology, although pulmonary embolus was suspected clinically.

Table 2. Related Adverse Events (n = 42 Patients)
 No. of Patients (%)
EventAny GradeGrade 3Grade 4
  1. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase.

Mucositis, functional42 (100)36 (86)0 (0)
Lymphopenia42 (100)27 (64)15 (36)
Leukopenia42 (100)24 (57)0 (0)
Pain, throat42 (100)8 (19)23 (55)
Fatigue42 (100)8 (19)0 (0)
Low hemoglobin42 (100)5 (12)0 (0)
Mucositis, clinical exam41 (98)9 (21)0 (0)
Nausea39 (93)9 (21)0 (0)
Weight loss38 (91)2 (5)0 (0)
Thrombocytopenia33 (79)0 (0)1 (2)
Vomiting32 (76)6 (14)0 (0)
ALT elevation30 (71)3 (7)0 (0)
Neutropenia28 (67)14 (33)4 (10)
Diarrhea24 (57)0 (0)0 (0)
Cough23 (55)0 (0)0 (0)
AST elevation18 (43)1 (2)1 (2)
Dermatitis in radiation field15 (36)1 (2)0 (0)
Pain, oral cavity14 (33)4 (9)3 (7)
Alkaline phosphatase elevation13 (31)0 (0)0 (0)
Hypertension13 (31)0 (0)0 (0)
Headache8 (19)1 (2)1 (2)
Febrile neutropenia7 (17)7 (17)0 (0)
Venous thrombosis/embolism3 (7)3 (7)0 (0)
Syncope3 (7)2 (5)1 (2)
Atrial fibrillation2 (5)1 (2)1 (2)

Patient 1 experienced a grade 4 pulmonary hemorrhage during maintenance bevacizumab treatment after the completion of chemoradiation. This patient did not have lung metastases, and it was believed that the event was related to pneumonia and bevacizumab. This patient remained without evidence of cancer at the time of this writing. After this grade 4 event in Patient 1, the protocol was amended, and no further patients received maintenance bevacizumab after chemoradiation (see Materials and Methods, above).


PFS and OS are illustrated in Figures 2 and 3, respectively. The median follow-up was 31.8 months as of April 30, 2011. The 2-year PFS rate was 75.9% (95% confidence interval [CI], 63.9%-90.1%), and the 2-year OS rate was is 88% (95% CI, 78.6%-98.4%).

Figure 2.

This Kaplan-Meier plot illustrates progression-free survival (solid line) with the 95% confidence interval (dotted lines).

Figure 3.

This Kaplan-Meier plot illustrates overall survival (solid line) with the 95% confidence interval (dotted lines).

Seven patients have developed recurrent disease. The first sites of failure were distant metastases for 5 patients (including lung in 3 patients; bone in 2 patients; and liver, stomach, peritoneum, and abdominal lymph node in 1 patient each). For 1 patient, disease recurrence was discovered concurrently in the regional neck lymph nodes and the lung. For 1 patient, the initial recurrence was at the site of the primary oropharyngeal tumor. Locoregional failure was suspected in 1 other patient but was not proven histologically. This patient was a man aged 54 years with a base of tongue squamous cell carcinoma (T4aN2cM0) who presented at an outside hospital with massive oropharyngeal hemorrhage approximately 8 months after completing study treatment, and he died of complications of that event. Locally recurrent tumor was suspected clinically, but biopsies taken at the outside hospital were negative for recurrent disease.

Late Functional Outcomes:

A man aged 68 years who had a base of tongue squamous cell carcinoma (T3N2cM0) and type 2 diabetes mellitus had delayed healing of an ulcer at the primary tumor site that required surgical debridement 7 months after completion of chemoradiation. Now, >4 years status postchemoradiation, he remains without evidence of disease and tolerates a full diet.

Post-treatment speech and swallowing function was evaluated with the Performance Status Scale for Head and Neck Cancer Patients (PSS-HN) among 32 patients.18 The median interval between completion of radiation therapy and the initial post-treatment HN-PSS assessment for 32 patients was 5.6 months (range, 1.3-26 months). Post-treatment HN-PSS results are provided in Table 3. Median scores for eating, speech, and diet were 100, 100, and 95, respectively. For 10 patients, HN-PSS follow-up data were not obtained because of recurrent disease or death (n = 6), lost to follow-up (n = 3), or data not obtained by investigator (n = 1).

Table 3. Performance Status Scale for Head and Neck Cancer Patients: Results After Chemoradiation (N = 32 Patients)
 PSS-NH Score 
FunctionMedianRangeNo. of Patients With a Score of 100 (%)
  1. Abbreviations: PSS-HN, Performance Status Scale for Head and Neck Cancer Patients.

Eating1000-10024/32 (75)
Speech10050-10027/32 (84)
Diet950-10016/32 (50)


This phase 2 study demonstrated the feasibility and encouraging efficacy of the addition of bevacizumab to the widely accepted regimen of high-dose cisplatin every 3 weeks given concurrently with IMRT for patients with locally and/or regionally advanced HNSCC. The 2-year PFS rate, which was the primary endpoint of the study, was 75.9%. Adverse events associated with the study regimen were similar to those described in prior studies of concurrent radiation and high-dose cisplatin for this patient population, although myelosuppression appeared to be increased compared with prior studies of cisplatin monotherapy.

At the time this study was designed, there were concerns about the possible risk of hemorrhage with bevacizumab among patients with squamous cell carcinomas.12 Our results indicate that the risk of tumor hemorrhage with bevacizumab-based combined modality therapy for HNSCC is low in appropriately selected patients. It is important to note the extensive exclusion criteria used in this study to minimize risk of bleeding complications. For example, the study excluded patients who had encasement of the carotid artery and patients on full-dose aspirin or therapeutic anticoagulation. Further clinical research would be needed to determine whether bevacizumab may be administered safely in an HNSCC population not subjected to these extensive exclusion criteria.

There were 2 deaths within 90 days of completing the study treatment (1 death from aspiration pneumonia and 1 sudden death of unclear etiology). The reported rates of treatment-related death among patients with HNSCC who receive cisplatin-based chemoradiation range from 1% to 5%. The combined results from 4 trials indicate that the rate of acute treatment-related death from standard cisplatin-based chemoradiation for locally advanced HNSCC is approximately 4%.1-3, 19 The results from the current study do not suggest that the addition of bevacizumab increases the risks of treatment-related death associated with chemoradiation. The results from Radiation Therapy Oncology Group (RTOG) trial RTOG 0615, a phase 2 study of the addition of bevacizumab to standard chemoradiation for locoregionally advanced nasopharyngeal cancer, are consistent with the safety experience in this study.20

We did not observe any clear evidence of increased local toxicity. The high incidence of acute functional mucositis grade ≥3 should be viewed in the context of the protocol requirement that all patients had placement of PEG tube before treatment. CTCAE version 3.0 defines grade 3 functional mucositis as the inability to adequately aliment or hydrate orally, and PEG tube use typically was scored as grade 3 functional mucositis during the course of treatment for the patients in this study. The incidence of grade ≥3 clinical mucositis was much lower (21%), indicating that bevacizumab did not appear to increase the acute local toxicities of chemoradiation. Post-treatment HN-PSS scores indicate that long-term recovery of speech and swallowing function was experienced by most patients in this study.

The incidence of grade ≥3 leukopenia (57%) was slightly higher than the 40% to 47% rate observed in randomized phase 3 studies of radiation plus bolus cisplatin in patients with HNSCC.1, 2 Although it is not possible to draw definitive conclusions based on comparisons with phase 3 studies, these results suggest that the addition of bevacizumab in this study may modestly intensify cisplatin-related leukopenia. This is consistent with a previous report that the myelosuppressive effects of cytotoxic chemotherapy appeared to have been increased by the addition of bevacizumab to the FHX regimen for patients with head and neck cancer.11 To reduce the potential risk of cumulative myelosuppression with 3 cycles of bolus cisplatin plus bevacizumab given concurrently with radiation therapy, it may be prudent to limit future studies to 2 cycles of concurrent bolus cisplatin plus bevacizumab in this patient population. We are conducting a follow-up combined-modality study for patients with stage III through IVB HNSCC that features only 2 cycles of cisplatin plus bevacizumab with concurrent radiation therapy. Patients also receive weekly, concurrent cetuximab in that follow-up study.21

The current study was designed for patients with locally or regionally advanced HNSCC of any subsite. However, in reality, 93% of patients on this study had oropharyngeal primary tumors. The high incidence of HPV-positive tumors and the tobacco histories (half of the patients were never-smokers or light smokers [≤1 pack-year]) in this study underscore the changing demographics of HNSCC. Recent reports have identified a gradual shift over 20 to 30 years toward an increasing prevalence of HPV-positive oropharyngeal squamous cell carcinoma and a decrease in the incidence of tobacco-related HNSCC in the United States and Western Europe.22-24 This change has dramatic implications with regard to prognosis. Data from RTOG 0129, a phase 3 study in which patients with locally and/or regionally advanced HNSCC received cisplatin concurrently with radiation therapy, demonstrated that 3-year OS among patients with HPV-positive oropharyngeal squamous cell carcinoma was superior to that of patients with HPV-negative oropharyngeal squamous cell carcinoma.25 These associations were not established at the time the current study was designed and limit any attempt to compare efficacy data from the current study with the results in historical controls.

A randomized phase 2 study of FHX with or without bevacizumab for patients with head and neck cancer was closed early to because of an increased locoregional failure rate in the experimental arm.26 The addition of bevacizumab to FHX required dose reductions on both 5-FU and hydroxyurea compared with standard FHX.11, 26 The reduced doses of cytotoxic chemotherapy in the FHX plus bevacizumab regimen could be causal with regard to the observed increased incidence of locoregional failure, although the small sample size (26 patients) in the phase 2 study precludes definitive conclusions regarding the efficacy of modified FHX plus bevacizumab.26 We did not observe any evidence of decreased efficacy with the addition of bevacizumab to bolus cisplatin given with daily IMRT in the current study.

This phase 2 study demonstrates that the addition of bevacizumab to bolus cisplatin administered concurrently with IMRT is feasible and yields encouraging efficacy among patients with stage III/IVB HNSCC. The incidence of grade ≥3 leukopenia appears to have been modestly higher than that observed in prior studies of cisplatin plus radiotherapy in this patient population. Although the study met its predetermined efficacy endpoint for 2-year PFS, comparisons with historic controls are difficult, because earlier studies typically did not contain the abundant representation of favorable-prognosis, HPV-positive oropharyngeal tumors that occurred in the current study. In view of the tolerability of this bevacizumab and the potential for bevacizumab to improve efficacy as a radiosensitizer and chemosensitizer, further prospective studies with bevacizumab in this patient population are ongoing.


This work was supported with funds from Genentech/Roche.


The authors made no disclosures.