Multicenter, prospective, phase II study of maintenance bevacizumab for children and adults with NF2-related schwannomatosis and progressive vestibular schwannoma

Abstract

Background

Prospective data on maintenance therapy with bevacizumab for persons with NF2-related schwannomatosis (NF2-SWN) is lacking. In this prospective multicenter phase II study, we evaluated the efficacy, safety, and tolerability of bevacizumab for maintenance therapy in children and adults with NF2-SWN and hearing loss due to vestibular schwannomas (VS).

Methods

Following induction therapy, participants received bevacizumab 5 mg/kg every 3 weeks for 18 months. Participants were monitored for changes in hearing, tumor size, and quality of life (QOL), and for adverse events. Hearing loss was defined as a statistically significant decline in word recognition score (WRS) or pure-tone average compared to the study baseline; tumor growth was defined as >20% increase in volume compared to baseline.

Results

Twenty participants with NF2-SWN (median age 23.5 years; range, 12.5–62.5 years) with hearing loss in the target ear (median WRS 70%, range 2%–94%) received maintenance bevacizumab. Freedom from hearing loss in the target ear was 95% after 48 weeks, 89% after 72 weeks, and 70% after 98 weeks. Freedom from tumor growth in the target VS was 94% after 48 weeks, 89% after 72 weeks, and 89% after 98 weeks. NF2-related QOL remained stable for 98 weeks whereas tinnitus-related distress decreased. Maintenance bevacizumab was well tolerated, with 3 participants (15%) discontinuing treatment due to adverse events.

Conclusions

Maintenance bevacizumab (5 mg/kg every 3 weeks) is associated with high rates of hearing and tumor stability during 18 months of follow-up. No new unexpected adverse events related to bevacizumab were identified in this population.

NF2-related schwannomatosis (NF2-SWN, formerly known as neurofibromatosis type 2¹) is a rare tumor predisposition syndrome characterized by the presence of multiple benign nervous system tumors including schwannomas, meningiomas, and ependymomas. Hearing loss due to vestibular schwannomas (VS) is nearly universal for persons with NF2-SWN. In addition, VS cause tinnitus, vertigo, brainstem compression, and lower cranial nerve dysfunction which independently and collectively impair quality of life (QOL).^2,3 Hearing improvement and tumor shrinkage occur in 30%–40% of persons with NF2-SWN and progressive VS treated with bevacizumab, a humanized antibody that inhibits vascular endothelial growth factor (VEGF).^4–9 Discontinuation of treatment is associated with a gradual decline in hearing and an increase in tumor size, indicating that long-term maintenance therapy is required to maintain clinical benefit. However, prolonged treatment with bevacizumab is associated with clinically relevant adverse events, including hypertension, proteinuria, menstrual irregularities, and delayed wound healing.¹⁰ Determining the activity and safety of maintenance bevacizumab is important for persons with NF2-SWN who are typically young at treatment initiation and are likely to require prolonged treatment.

Previous clinical trials of NF2-SWN patients have provided bevacizumab at 5 mg/kg every 2 weeks (or equivalent).^5,11 This clinical trial (NCT01767792) was designed to address 2 questions regarding treatment with bevacizumab. Induction treatment was provided at 10 mg/kg every 2 weeks to determine whether dose intensification could improve hearing responses. In contrast, maintenance therapy was provided at 5 mg/kg every 3 weeks to determine whether dose reduction could minimize toxicity without sacrificing efficacy. Participants who experienced a statistically significant hearing decline during maintenance therapy compared with the study baseline were transitioned to bevacizumab 10 mg/kg every 2 weeks to determine if the hearing decline was related to underdosing of bevacizumab. Participants who experienced further hearing decline on bevacizumab 10 mg/kg every 2 weeks discontinued treatment. We previously reported the safety, efficacy, and biomarker results for induction therapy.¹¹ We now report the efficacy and safety of bevacizumab for participants during maintenance therapy.

Materials and Methods

Participants, study design, and treatment

The details of this multi-institution, open-label phase II trial (NF104) run by the Department of Defense (DOD)-funded Neurofibromatosis Clinical Trial Consortium have been published.¹¹ Key inclusion criteria included age ≥6 years, clinical diagnosis of NF2,^12–14 clinical progression of VS during the prior 18 months (defined as either a statistically significant decline in WRS or tumor volume growth of ≥20%), baseline word recognition score (WRS) between 6% and 84% in the target ear, and at least 1 VS ≥ 0.4 ml on volumetric analysis of MRI. Key exclusion criteria included prior antiangiogenic therapy, medical conditions incompatible with bevacizumab treatment, and tumors not amenable to volumetric MRI analysis. Participants with stable or improved WRS after 6 months of induction treatment (10 mg/kg every 2 weeks) were subsequently treated with maintenance bevacizumab 5 mg/kg every 3 weeks for 18 months. (Figure 1). The trial was approved by the Human Research Protection Office at the DOD and by site Institutional Review Boards. Bevacizumab was supplied by Genentech. All participants or their legal guardians provided informed consent and pediatric participants provided assent.

Assessments

The target VS was assigned by the treating physician. Contralateral VS, if present, were monitored in similar fashion to target VS. Audiology was performed at baseline, every 12 weeks during treatment, and at the end of treatment. WRS, a measure of hearing intelligibility, was assessed using a 50-word list of monosyllable words delivered via standardized methodology.^15,16 The 95% critical-difference table was used to determine hearing responses, including hearing improvement, stable hearing, or hearing decline, as previously reported.^15,17 Pure-tone average (PTA) was used to assess hearing thresholds, a measure of hearing sensitivity; PTA is measured in decibels with higher levels indicating a greater degree of hearing loss. An increase in PTA of 12 dB or greater was defined as a hearing decline.

Brain MRI was performed at the study baseline, every 12 weeks during treatment, and at the end of treatment. MRI scans included postcontrast imaging through the internal auditory canal (slice thickness ≤3 mm) and the entire brain. Volumetric analysis was performed centrally by independent radiologists as previously described.¹⁸ Changes in VS volumes compared with the study baseline were determined for the target and, when present, for contralateral VS ≥ 0.4 ml at baseline. Radiographic response (RR) was defined as ≥20% decrease in tumor volume, progressive disease (PD) as ≥20% increase in tumor volume, and stable disease (SD) as all other results.

The NF2 Impact on Quality of Life (NFTI-QOL) questionnaire was used to measure NF2-specific QOL.¹⁹ This validated questionnaire assesses 8 domains including hearing, dizziness and balance, facial weakness, sight, mobility and walking, role and outlook on life, pain, and anxiety and depression. Participants score each domain from 0 (not present) to 3 (stops usual activities). The responses are summed (ranging from 0 to 24), with higher scores indicating worse disease-specific QOL. The psychological distress caused by tinnitus was evaluated using the Tinnitus Reaction Questionnaire (TRQ), as previously described.²⁰ The TRQ consists of 26 statements such as “my tinnitus has made me unhappy” which participants rate from 0 (not at all) to 4 (almost all of the time). The responses are summed (ranging from 0 to 104), with higher scores indicating more distress related to tinnitus. Participants completed NFTI-QoL and TRQ every 6 months during maintenance therapy. The minimal clinical important difference (MCID) in these scores was defined as previously described.²⁰

Adverse events (AEs) were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0. Serum FSH levels were measured every 3 months in menstruating women to screen for possible premature ovarian insufficiency (POI). Height and X-rays of the tibial growth plate were monitored in pediatric patients.

Statistical analysis

Participant and disease characteristics at the start of maintenance are presented with standard descriptive summaries. The Kaplan–Meier method was used to describe the proportion of participants with freedom from hearing loss and from tumor growth. Survival analyses were performed from the start of maintenance therapy and from the start of treatment. The QOL data are presented as means with standard deviations. All Pvalues are reported as 2-sided and significance level = 0.05. All analyses were conducted using SAS software (version 9.4, SAS Institute).

Results

Participants and study treatment

Twenty participants (13 female participants) with a median age of 23.5 years (range, 12.5–62.5) were treated with maintenance bevacizumab. The maintenance cohort included 6 pediatric participants (<21 years of age) and 14 adult participants. Median WRS at the start of maintenance therapy was 70% (range, 2%–94%) in the target ear. Nine participants (45%) were at risk for complete deafness due to pre-existing anacusis in the contralateral ear. Sixteen participants (75%) completed all 18 months of maintenance treatment. Four participants discontinued treatment: at week 28 (adult participant due to nonhealing perianal abscess), at week 82 (pediatric participant due to patient/family choice for grade 1/2 toxicity), at week 83 (adult participant due to progressive hearing loss), and at week 85 (adult participant due to clinician choice for grade 2 proteinuria).

Hearing function

Hearing decline compared with baseline, as measured by WRS, was documented in 3 target ears (15%) during maintenance therapy, including 1 adult and 2 pediatric participants (Figure 2). The adult was transitioned to high-dose bevacizumab (10 mg/kg every 2 weeks) per the protocol and experienced hearing improvement to the stable range. The 2 pediatric patients experienced hearing loss at the final on-study visit (week 98) and therefore were not eligible for treatment at higher of dose bevacizumab on the protocol. Among 9 non-target ears, 1 pediatric patient experienced hearing loss (Figure 2). Hearing loss during maintenance was more common in pediatric participants (3/11 ears, 27%) than in adult patients (1/18 ears, 6%) although this did not reach statistical significance (P = 0.054). Freedom from hearing loss in the target ear during maintenance therapy was 95% after 48 weeks, 89% after 72 weeks, and 70% after 98 weeks; in the non-target ear, the rates were 100% after 48 weeks, 100% after 72 weeks, and 71% after 98 weeks.

Hearing decline as measured by PTA was documented in 2 target ears (10%) during maintenance therapy (Figure 3) but did not coincide with the hearing loss measured by WRS. The adult participant experienced hearing decline on 1 evaluation only (week 48); subsequent PTA values returned to the stable range without intervention. The pediatric patient also experienced hearing decline at the final-on-study visit (week 98) and was not tested thereafter. Among 9 non-target ears, 1 pediatric and 1 adult participant experienced hearing loss by PTA (Figure 3). As with target ears, hearing loss was documented at a single time point. Freedom from hearing loss in the target ear as measured by PTA was 89% after 48 weeks, 89% after 72 weeks, and 83% after 98 weeks; in the non-target ear, the rates were 100% after 48 weeks, 100% after 72 weeks, and 89% after 98 weeks.

Tumor size

The median baseline volume at the start of maintenance was 2.8 cc (range 0.8–16.9 cc) in the target VS and 1.5 cc (range 0.3–12.8 cc) in the non-target VS. A total of 37 VS (20 targets and 17 contralaterals) were evaluable for tumor growth. Progressive disease during maintenance therapy occurred in 2/20 (10%) target and 4/17 (24%) non-target ears (Figure 4). In total, 6/37 (16%) VS progressed during maintenance therapy. PD was significantly more common in VS of pediatric participants (5/10 VS, 50%) than in VS of adult patients (1/27 VS, 4%; P < 0.001). Freedom from tumor growth in the target VS was 94% after 48 weeks, 89% after 72 weeks, and 89% after 98 weeks; in the non-target VS, the rates were 80% after 48 weeks, 73% after 72 weeks, and 50% after 98 weeks.

Safety and tolerability during maintenance

There were 311 treatment-emergent adverse events that occurred in 20 participants (100%) (Table 1). Treatment-related adverse events of any grade and grade 3 occurred in 19 (95%) and 7 (35%) participants, respectively. No grade 4 or 5 AEs were reported. Adverse events reported in ≥10% of participants included hypertension (65%); irregular menstruation (46%); fatigue (30%); headache (20%); epistaxis (20%); proteinuria and constipation (15%); and arthralgia, nausea, abdominal pain, gum bleeding, diarrhea, rectal hemorrhage, and dry skin (10% each). Grade 3 treatment-emergent AEs included irregular menstruation, premature menopause, headache, mucositis, perianal abscess, vomiting, and dental wound. No patients experienced thromboembolic disease. Five/13 women (46%) with experienced grade 1–3 irregular menstruation, 3/13 (23%) experienced elevated FSH, and 1 experienced grade 3 premature menopause. All women elected to continue treatment after a discussion of potential risks.

Quality of life during bevacizumab treatment

The NF2 Impact on Quality of Life (NFTI-QOL) was completed by 20, 20, 17, 17, and 14 participants at 0, 6, 12, 18, and 24 months, respectively. The average summary score at the study baseline was 7.7 (SD 3.3). The mean NFTI-QOL score dropped to 7.1 (SD 3.7) at 6 months (initiation of maintenance therapy) and remained near this level throughout maintenance treatment (Figure 5). For adults, the average summary score dropped from 7.8 (SD 3.5) at baseline to 7.6 at 6 months (start of maintenance therapy) and then to 6.7–7 throughout maintenance therapy. In contrast, for 6 pediatric participants, the baseline NFTI-QOL score was 7.3 (SD 3.2) and dropped to 6.0 at 6 months but then increased above baseline during maintenance therapy (Figure 5). At no time point, did mean QOL scores exceed the MCID of 2, the threshold for clinically important improvement in QOL for the group.

Tinnitus reaction questionnaire (TRQ) was completed by 20, 20, 20, 17, 17, and 14 participants at 0, 3, 6, 12, 18, and 24 months, respectively. The average score at the study baseline was 22 (SD 18.7) which decreased significantly between 6 (initiation of maintenance therapy) and 12 months of treatment (Figure 5; P < 0.05) but did not exceed the MCID of 10 at any time point. The mean TRQ score increased between 12 and 24 months of treatment but remained arithmetically below the baseline. In contrast to the NFTI-QOL scores, the reduction in tinnitus-related distress was experienced by both adult and pediatric participants (Figure 5).

Discussion

Bevacizumab is recommended for the treatment of VS in NF2-SWN patients by the European Association of Neuro-Oncology.²¹ Two doses of bevacizumab have been approved for cancer treatment by the Food and Drug Administration and the European Medicines Agency: 2.5 mg/kg/week dose equivalent and 5.0 mg/kg/week dose equivalent.¹⁰ However, guidance on long-term dosing for NF2-SWN patients with benign tumors is lacking, and studies of vascular normalization with anti-VEGF drugs suggest that lower doses may improve outcomes.^22,23 In this phase II trial involving participants with NF2-SWN and progressive vestibular schwannoma, treatment with bevacizumab 5 mg/kg every 3 weeks (1.67 mg/kg/week dose equivalent) after induction therapy was effective in preventing hearing decline and tumor growth, in maintaining NF2-related QOL, and in reducing distress related to tinnitus.

Our findings support those of previously reported retrospective studies of bevacizumab in persons with NF2-SWN. Small retrospective studies of persons receiving low-dose bevacizumab during maintenance therapy (5 mg/kg every 3 weeks or 2.5–5 mg/kg every 4 weeks) report freedom from tumor growth in 90%–100% at 1 year and in 63%–88% at 2 years,^8,24,25 which is equivalent to the 94% and 89% reported in our study and compares favorably with rates of 69% and 36% in newly diagnosed patients enrolled in a natural history study.³ To date, the minimum effective dose of bevacizumab for maintenance therapy has not been determined, and clinicians should choose among the various dosing regimens to best fit an individual patient’s needs.

A posthoc analysis of clinical outcomes by age revealed important differences. More pediatric participants experienced hearing loss (27% of eligible ears) or tumor growth (50% of VS) than adult participants (6% of eligible ears and 4% of VS) during maintenance therapy. A similar trend for reduced efficacy of bevacizumab for pediatric patients was noted during induction therapy and in retrospective studies of bevacizumab for pediatric patients.^8,11 Our findings are also aligned with those reported by a nationally coordinated health care system in the United Kingdom⁸; by pediatric oncology centers in France⁷; and by 6 international pediatric institutions²⁶ suggesting that pediatric patients benefit less from bevacizumab than adults. The basis for this difference is not yet known. However, age at first symptom is strongly associated with disease severity, and individuals diagnosed under age of 16 years typically exhibit a more severe phenotype with early hearing loss and mortality.²⁷

Finally, we document improvement in NF2-related QOL and in tinnitus-related distress with bevacizumab treatment that corresponds with previous reports,^8,20 although all studies have been limited by low-patient numbers. In contrast to previous studies,⁸ our results suggest that some ears that experience hearing decline on low-dose bevacizumab can be salvaged by an increase in bevacizumab. This finding highlights the importance of individualizing treatment and considering dose intensification for those who have exhibited stable or improved hearing on higher doses of bevacizumab previously.

Bevacizumab was well tolerated during 2 years of treatment. Only 2 participants chose to stop treatment due to grade 1 or 2 adverse events, and 1 was stopped by the treating physician for grade 3 toxicity. The rate of hypertension in this study (65%) was higher than reported in a recent meta-analysis of 14 studies for persons with NF2-SWN (29%),²⁸ and may reflect to the prospective nature of the trial, the duration of treatment, or other unmeasured differences in the cohort. In contrast, the rate of proteinuria (15%) was somewhat lower than in the meta-analysis (30%) whereas the rates of menstrual irregularities were similar. No new safety signals for bevacizumab were identified during maintenance therapy.

Our trial had several strengths, including a prospective trial design that posed independent questions about induction therapy and maintenance therapy, enrollment of participants across multiple dedicated NF clinics, a relatively long treatment period with bevacizumab, and a primary outcome (hearing) which is directly related to QOL for people living with NF2-SWN.

This trial also has limitations. First, given the rarity of NF2-SWN and previous reports of treatment efficacy, a double-blind trial design was not possible. Second, the analysis by age was a posthoc assessment which limits the ability to make conclusions about the effect of age on response to bevacizumab. Third, the trial was limited to participants with progressive VS enrolled at specialty centers, potentially limiting the generalizability of these findings. Despite these limitations, our results agree with the previous retrospective case series.

For NF2-SWN patients with progressive VS, treatment with low-dose bevacizumab as maintenance therapy following higher dose induction therapy is associated with high rates of hearing preservation and low rates of tumor progression. Additional studies are required to determine the minimum effect dose of bevacizumab for maintenance and whether this treatment approach may be useful to maintain clinical response and minimize side effects beyond 2 years.

Funding

Dana-Farber/Harvard Cancer Center is supported in part by NCI Cancer Center Support Grant # NIH 5 P30 CA06516. The Department of Defense (DOD) Neurofibromatosis Clinical Trials Consortium is supported in part by Department of Defense grant # W81XWH-12-1-0155.

Acknowledgments

Research support for this study was provided by Genentech. We thank Juliette Southworth, Lynn Merritt, Steve Powell, Vivien Philips, and Karen Cole-Plourde for their assistance. We thank Dana-Farber/Harvard Cancer Center in Boston, MA, for the use of the Tumor Imaging Metrics Core, which provided centralized imaging service.

Data availability: The data presented in the manuscript is not currently available in a public repository. For data inquiries, please reach out to the NF Clinical Trials Consortium.

Funding: Research support for this manuscript was provided by Genentech and the Department of Defense.

Conflict of interest

Scott Plotkin: co-founder of NF2 Therapeutics; consultant for Akouos. Jeffrey Allen: No conflicts to report. Girish Dhall: No conflicts to report. Jian Campian: research support from Merck, NeoimmuneTech, Incyte, and GI Innovation. D. Wade Clapp: No conflicts to report. Michael Fisher: Advisory Board for AstraZeneca and Springworks; research support from AstraZeneca, Array Biopharma, and Exelixis. Rakesh Jain: Consultant fees from BMS, Elpis, Innocoll, SPARC, SynDevRx; owns equity in Accurius, Enlight, SynDevRx; Serves on the Board of Trustees of Tekla Healthcare Investors, Tekla Life Sciences Investors, Tekla Healthcare Opportunities Fund, Tekla World Healthcare Fund, and received Research Grants from Boehringer Ingelheim and Sanofi. James Tonsgard: No conflicts to report. Nicole Ullrich: No conflicts to report. Coretta Thomas: No conflicts to report. Lloyd Edwards: No conflicts to report. Bruce Korf: Medical Advisory boards: Genome Medical, Infixion, Healx, and SpringWorks, Recursion. Roger Packer: No conflicts to report. Matthias Karajannis: consultant for CereXis, Recursion. Jaishri Blakeley: Medical Advisory Board for SpringWorks.

Authorship

Study concept and design: Scott Plotkin, Jeffrey Allen, Michael Fisher, Nicole Ullrich, Bruce Korf, Roger Packer, Matthias Karajannis, and Jaishri Blakeley. Data collection: Scott Plotkin, Jeffrey Allen, Girish Dhall, Jian Campian, D. Wade Clapp, Michael Fisher, James Tonsgard, Nicole Ullrich, Roger Packer, Matthias Karajannis, and Jaishri Blakeley. Analysis and interpretation of data: Scott Plotkin, Jeffrey Allen, Michael Fisher, Rakesh Jain, Nicole Ullrich, Coretta Thomas, Lloyd Edwards, Matthias Karajannis, and Jaishri Blakeley. Drafting of the manuscript: Scott Plotkin, Coretta Thomas, Lloyd Edwards, and Jaishri Blakeley. Revising of the manuscript: All authors.

References

Author notes