Phase II trial of pegylated interferon alfa-2b in young patients with neurofibromatosis type 1 and unresectable plexiform neurofibromas

Abstract

Background.

There is no proven medical therapy for plexiform neurofibromas (PNs). We undertook a phase II trial of pegylated interferon (PI) to evaluate response and time to progression (TTP).

Methods.

PI was administered as a subcutaneous injection to patients with neurofibromatosis type 1‒related PN, stratified by the presence of symptoms (asymptomatic: stratum 1, symptomatic: stratum 2) or documented imaging progression (stratum 3). Patients in strata 1 and 2 received PI for up to one year if stable, 2 years for those with clinical (stratum 2) or imaging response (≥20% decrease in volume). Patients on stratum 3 continued PI until progression. PI was considered active in stratum 3 if TTP doubled compared with the placebo arm of a previous randomized trial using tipifarnib.

Results.

Enrolled were 82 evaluable patients (median age 10 y; range 1.6 to 21.4). Fatigue and/or worsening of behavioral issues were the most common toxicities requiring dose modification. Across all strata, imaging responses were seen in 4 patients (5%). Three of 26 symptomatic patients on stratum 2 met the criteria for clinical response without corresponding imaging changes. In stratum 3, median TTP was 29.4 months versus 11.8 for the placebo arm of the previous trial (P=.031). The slope of tumor growth on PI slowed significantly compared with the slope before starting PI (P=.044).

Conclusions.

In patients with active PN, PI results in more than doubling of the TTP compared with placebo. Imaging changes in symptomatic patients were not associated with changes in clinical status.

Neurofibromatosis type 1 (NF1) is a common autosomal dominant neurocutaneous disorder that carries a high risk of tumor development, particularly in the central and peripheral nervous system.¹ Plexiform neurofibromas (PN) are one of several types of neurofibromas that frequently occur in patients with NF1. They consist of a network of neoplastic Schwann cells accompanied by a variable amount of cellular and noncellular components, including fibroblasts, perineural cells, and mast cells that grow along the length of nerves, often involving multiple nerve fascicles, branches, and plexi.² They are clinically evident in approximately 25% of patients with NF1,³ although imaging reveals an incidence as high as 50%.⁴ Complete surgical resection is virtually impossible as the tumors typically infiltrate adjacent normal tissue. As they grow, they can become disfiguring, painful, disabling, or even life threatening as they compress vital structures.^5,6 Standard chemotherapy is not effective and there is currently no proven medical treatment despite various clinical trials evaluating potential therapies.^7–9

Type 1 interferons, such as alpha-interferon, are cytokines predominantly produced by leukocytes in response to viral infections and are known to have antiproliferative, antiviral, and immunoregulatory activities.¹⁰ In addition, they have both direct^11,12 and indirect antitumor activity. The indirect mechanisms include the induction of immunomodulatory factors, such as the activation of cytotoxic T-lymphocytes, natural killer cells, and monocytes,¹³ and the induction of increased cell surface expression of class I major histocompatibility complex antigens¹⁴ and nonimmunomodulatory factors, such as inhibition of transcription and secretion of anti-angiogenic factors.¹⁵ A study evaluating short acting interferon alpha-2a in patients with PN showed a high rate of prolonged radiographic stability by standard imaging as well as clinical improvement in approximately 20%.¹⁶ The pegylated form of interferon (PI) has a significantly prolonged plasma half-life, allowing protracted activity and less frequent administration. A phase I study of PI in pediatric patients with unresectable PN showed both clinical and imaging responses by volumetric MRI analysis and established the recommended phase II dose of 1.0 mcg/kg/wk given by subcutaneous injection.¹⁷ We therefore undertook a phase II trial evaluating the imaging (stratum 1) and/or clinical (stratum 2) response rates in patients with unresectable PN. Stratum 3 evaluated the impact of PI on growth rate in patients with documented progression on imaging at trial entry by comparing time to progression (TTP) with the actuarial curve of the placebo arm of the National Cancer Institute–coordinated tipifarnib trial (NCT00021541).⁹

Patients and Methods

The study was approved by the institutional review board of all participating institutions. Written informed consent was obtained from patients aged ≥18 years or from parents/legal guardians of children aged <18 years, with child assent when appropriate according to individual institutional policies. The trial was registered with www.ClinicalTrials.gov, identification number NCT00678951.

Eligibility

Patients ≥18 months to 21 years of age with at least one inoperable PN that had the potential to cause significant morbidity were eligible. Histologic confirmation of PN was not required in the presence of consistent clinical and radiographic findings. However, if any clinical observation or scan suggested possible malignant transformation, the tumor was to be biopsied prior to enrollment. Patients had to meet the diagnosis of NF1 as defined by the NIH Consensus Conference¹⁸ or undergo a biopsy to confirm the pathology of PN. Patients had to have recovered from all toxic effects of previous therapy, a life expectancy of at least 12 months and an Eastern Cooperative Oncology Group performance score of 0, 1, or 2. Adequate bone marrow (absolute neutrophil count ≧1500/µL, hemoglobin >10 gm/dL, platelet count ≧ 100,000/µL), renal (normal serum creatinine for age or a creatinine clearance ≥70mL/min/1.73 m²) and hepatic (total bilirubin < 1.5 × normal and serum glutamic pyruvic transaminase <2 × upper limit of normal) function was required. Baseline laboratory and imaging studies, including an MRI scan of the target lesion with coronal and axial short inversion time inversion recovery (STIR) images of the entire PN as specified by the protocol were obtained within 4 weeks of study entry. Baseline physical exam was typically done on the date of enrollment when parents were taught to give injections. Patients with orbital PN underwent a protocol-specified baseline ophthalmologic evaluation. Patients with pain associated with the target PN were required to complete a pain medication diary for at least one week prior to study enrollment.

Exclusion criteria included any clinically significant unrelated systemic illness, cardiovascular disease or severe psychiatric condition, pregnancy or lactation, exposure to an investigational chemotherapy agent within 30 days, a visual pathway glioma requiring treatment, a history of a malignant peripheral nerve sheath tumor, or prior treatment with interferon.

Patients were entered on one of 3 strata (Table 1), depending on whether they had documented imaging or clinical progression during the 12 months preceding trial entry, an orbital location, or required daily pain medications. Patients in stratum 1 did not need to demonstrate loss of function or imaging progression but were required to have a PN with risk for significant morbidity. Criteria for stratum 2 eligibility included loss of function due to PN(s) as reflected by a decrease in the Karnofsky or Lansky score by ≥1 level within the year preceding trial entry or target PN-associated pain requiring daily pain medication. Patients with an orbital location were eligible if followed by an ophthalmologist familiar with the study guidelines. Imaging progression for eligibility on stratum 3 was the same as for the trial providing the historical controls⁹ and defined as the presence of a new PN on MRI or a ≥20% increase in volume, a ≥13% increase in the product of the 2 longest perpendicular diameters, or a ≥6% increase in the longest diameter of the target PN over the last 2 consecutive scans or approximately the last year prior to trial evaluation. The latter 2 are the mathematical calculations of the changes that correspond to a 20% increase in volume when a lesion is measured in either one or 2 dimensions. Documentation of growth by partial volume measurements was allowed if only a segment of the lesion was consistently imaged as long as the images had clear anatomic landmarks to ensure that the same area was reliably measured.

Treatment Plan

PI was administered as a weekly subcutaneous injection at a dose of 1.0 mcg/kg/wk. Acetaminophen (15mg/kg up to max dose of 1000mg) was given 30 minutes prior to the first dose of PI and then every 4–6 hours as needed. Patients on strata 1 and 2 could remain on study for up to 1 year unless they had a response, in which case they continued treatment for a maximum of 2 years. Patients on stratum 3 were to continue treatment until disease progression. Patients were removed from treatment if they developed disease progression, defined as ≥20% increase in tumor volume compared with baseline or clinical progression in conjunction with an increase in tumor volume, intolerable toxicity, or at the request of the patient or parent.

Monitoring/Evaluation Schedule

Patients were evaluated at week 6, then at months 4, 8, 12, 18, and 24 while on treatment and every 6 months thereafter until off study. Toxicity was evaluated, and complete blood counts and chemistries with liver function tests were performed at each visit and as indicated in the interim. Drug diaries to confirm compliance and pain medication diaries for patients on stratum 2 were reviewed at each visit. Patients with orbital tumors on stratum 2 underwent formal ophthalmologic evaluations, and those with pain had their pain medication diaries collected on the same schedule as the MRI scans.

MRI scans were performed at baseline and at months 4, 8, 12, 18, and 24 while on treatment and every 6 months thereafter until off study. Axial and coronal STIR images, on which PNs are bright compared with normal surrounding tissue, were obtained to encompass the entire PN using a slice thickness of 5–10mm with no skips between slices.

Determination of imaging response was undertaken centrally at the Pediatric Oncology Branch of the National Cancer Institute (NCI) using a validated method of semiautomated 3D MRI analysis to measure changes in PN volume.¹⁹ Tumor-containing areas were determined from STIR images on each MRI and the data were processed to assess the volume of the target PN(s), defined as the 1–3 most clinically relevant (ie, the PN[s] which had the greatest likelihood to cause clinical compromise). The volumetric analysis was used to determine response and disease progression by comparing the results either with the tumor volume in the pretreatment MRI or with the tumor volume at the time of best response. A response was defined as a ≥20% reduction in the sum of the volume of the “target PN,” confirmed by a follow-up MRI after ≥4 weeks. Disease progression was defined as a ≥20% increase in tumor volume.

Toxicity Evaluations and Dose Modifications

Toxicities were graded according to the NCI Common Toxicity Criteria (CTC version 4.0). PI was held for any drug-related grade 3 or 4 nonhematologic toxicity (NHT). Patients who developed grade 4 NHT, grade 3 hyperbilirubinemia, or grade 2 or higher cardiopulmonary toxicity were taken off study. Patients who experienced reversible grade 3 nonhematologic toxicity could restart therapy at the 0.6 mcg/kg/wk dose level provided toxicity resolved to grade 1 or better within 3 weeks. Patients who developed intolerable behavioral changes or persistent grade 2 or higher constitutional symptoms had the dose reduced to 0.6 mcg/kg/wk. Patients with other persistent (>7 days) grade 2 PI-related toxicities had subsequent doses withheld until the toxicity resolved to grade ≤1 and then restarted at the same dose level. If the grade 2 toxicity recurred, the PI dose was withheld again until the toxicity resolved to grade ≤1 and then reduced to 0.6 mcg/kg/wk. The dose was decreased to 0.3 mcg/kg/wk for any toxicity necessitating dose reduction at the 0.6 mcg/kg/wk dose level. Recurrent dose-limiting NHT following 2 dose reductions required discontinuation of drug. No subsequent dose escalations were permitted after dose reductions for toxicity.

Patients who developed grade 4 leukopenia or neutropenia or grade 3 or 4 thrombocytopenia had subsequent doses of PI held until the count(s) recovered to grade ≤1, at which point PI was restarted at a dose of 0.6 mcg/kg/wk. If the toxicity recurred, a second dose reduction to 0.3 mcg/kg/wk was allowed provided the toxicity resolved to grade ≤1 within 2 weeks. Further recurrence of dose-limiting hematologic toxicity resulted in permanent discontinuation of drug.

Statistical Considerations

The primary endpoint was imaging response for stratum 1, imaging and clinical response for stratum 2, and TTP for stratum 3.

A single stage design was used for strata 1 and 2, since determination of response might not be determined for more than 12 months, limiting the utility of a formal 2-stage design. In both strata 1 and 2, it was felt to be desirable to rule out a 5% response rate in favor of a more desirable, targeted response rate of 20%. By enrolling 27 patients into each stratum, there was 81% power to reject a 5% response rate in favor of a 20% response rate, using a one-sided 0.10 significance level exact binomial test with a 90% confidence interval about the observed response proportion. Therefore, if 4 or more responses were noted in 27 patients, this was considered desirable, as the lower one-sided 90% confidence interval bound on 4/27 is 6.6%. The trial also incorporated a stopping rule in which if there were no responses noted in the first 14 patients on a given stratum, accrual to that stratum would end as soon as this could be determined. Patients were considered inevaluable and replaced if they received 4 or fewer doses of PI.

Clinical response in stratum 2 was defined as a protocol-specified improvement in ophthalmologic evaluation, an improvement of at least one level in performance status (PS), ≥50 % decrease in the amount of pain medications required per week compared with baseline, or ability to change from a narcotic to a nonnarcotic analgesic, sustained for at least one month.

In the third stratum, the TTP actuarial curve was compared with that of patients enrolled on the placebo arm of the NCI trial evaluating tipifarnib in children with NF1-related PN.⁹ Based on a median TTP of 6 months in the placebo arm, which was estimated prior to knowing the results of the tipifarnib study, with 30 patients enrolled on both the placebo arm of the tipifarnib study and the present study, there would have been 81% power to detect a difference between the 2 arms, using a one-sided 0.05 alpha level log-rank test, assuming that the 2 arms had exponential TTP curves, with median TTP of 6 months and 12 months. It was assumed that accrual would be completed in 30 months on the present study with an additional 18 months of follow-up after the last patient had enrolled before making a final comparison between the arms. The actual median TTP for the placebo arm on the tipifarnib study is 11.8 months, but the power was never recalculated for the comparison with the present trial. Imaging response was a secondary endpoint.

TTP was estimated by the Kaplan–Meier method with a 2-tailed log-rank test used to determine the significance of the difference between 2 Kaplan–Meier curves. A Wilcoxon signed rank test was used to determine the significance of the change in slope of tumor growth after PI versus prior to PI. Spearman correlation analysis was used to assess the correlation between changes in tumor growth slopes and age. All P -values were 2 tailed.

Results

A total of 86 patients (median age 10.0 y, range 1.6–21.4 y for the 82 evaluable patients) were enrolled between December 2006 and April 2014. See Table 2 for the characteristics of the eligible/evaluable patients by stratum. Details of toxicity and response for each stratum are found in Table 3.

Stratum 1

Twenty-seven patients (median age 12.4 y, range 2.8–20.5 y) were enrolled on stratum 1. One patient came off study after the first several weeks due to an allergic reaction. One patient with significant total-body PN tumor burden developed a low-grade malignant peripheral nerve sheath tumor of the thigh 5 months after initiation of PI and came off study with stable disease (SD).

One patient had an unconfirmed imaging response (ie, follow-up MRI did not show a continued ≥20% decrease in volume), with a 22% decrease in volume at one year followed by slow progression over the next year, eventually coming off treatment due to progression after 2 years of treatment. Eighteen patients completed 12 courses or 1 year of treatment with SD and came off study according to protocol. Among the first 14 patients, 2 exhibited some improvement, including one with the unconfirmed imaging response, providing marginal evidence to not invoke the early stopping rule. Since fewer than 4 of 27 ultimately exhibited a response, the findings in this stratum do not rule out a 5% response rate.

Four patients withdrew consent during the course of treatment: 2 because they didn’t like the injections,one because of worsening depression, and another with SD after 8 months who did not feel that the PI was helping. Three patients required dose reductions for toxicity: 2 because of worsening behavioral problems and one because of fatigue. The latter patient subsequently came off study for worsening of baseline depression (grade 3).

Stratum 2

Twenty-nine patients (median age 9.4 y, range 1.7–21.1) were enrolled. Two patients were deemed inevaluable: one patient came off study within the first few weeks after a brain MRI showed a right cerebellar lesion that required surgical resection; the second patient died of a tracheostomy obstruction after receiving 2 doses of PI, felt to be unrelated to the PI. A third patient enrolled for pain was retrospectively considered ineligible because the patient was not receiving daily pain medications. Of the 26 eligible/evaluable patients, 14 were enrolled using the pain criteria, 8 for orbital locations, and 4 for a decline in PS. Their characteristics are presented according to their enrollment eligibility criteria in Table 2.

Two of 14 patients with pain (14%) and one of 4 with a decreasing PS met the criteria for clinical response (Table 3), without a corresponding decrease in volume, and have all received at least 24 months of treatment. One patient with an orbital PN had a confirmed imaging response and completed 24 months of treatment. Therefore, 4 of the 26 evaluable patients (15.4%; with exact lower one-sided 90% CI bound of 6.9%) had a clinical or imaging response, permitting us to rule out 5% as the response rate in this stratum.

Two patients came off study for PD after 8 and 12 cycles. An additional patient had significant worsening of arm paresthesias related to an axillary PN despite a 10% decrease in volume at 4 months and came off study by parental request.

Four patients enrolled for pain required dose reductions and/or came off study for worsening of preexisting behavioral issues or depression. In total, 9 patients (34.6%) required dose reductions, including 4 for grades 2–3 fatigue.

Stratum 3

Thirty patients (median age 7.1 y, range 1.6–17.6) were enrolled. One patient was ineligible due to lack of documented progression at baseline. Two patients had a ≥20% decrease in PN volume, one of which was followed by progression and thus unconfirmed. Eighteen patients developed PD while on treatment at a median of 18 months, range 8–36 months. Eight patients (27.6%) with SD withdrew consent and/or became noncompliant after 8 months to 4.5 years of treatment. Two patients continue to receive drug on study for 58+ and 63+ months.

One patient came off study for recurrent transaminitis despite 2 dose reductions. Five patients required dose reductions for grades 2–4 constitutional issues.

The slope of tumor growth on PI slowed significantly compared with the slope prior to starting PI (P=.044 by Wilcoxon signed rank test; see Fig. 1). There was no significant association between the changes in tumor growth slope and patient age (r=0.21; P=.37). There was no difference in age (P=.25) or tumor volume (P=.46) between the patients enrolled on stratum 3 and those on the placebo arm of the tipifarnib trial.⁹ Median TTP for the 29 eligible patients is 29.4 months versus 11.8 months for the placebo arm of the tipifarnib trial (P=.031, by 2-tailed log-rank test; see Fig. 2).

Discussion

NF1-associated PN can cause significant morbidity and even mortality, and effective treatment remains an area of unmet need. Over the years, many agents have been evaluated; although the results to date have largely been disappointing, progress has been made in the ability to appropriately assess the efficacy of newer agents for these tumors. The development of a validated semiautomated method¹⁹ to evaluate serial volumetric tumor measurements has allowed more accurate and sensitive assessments of imaging response as well as growth in tumors like PN, whose imaging characteristics are not amenable to the standard means of using 2-dimensional cross-sectional diameters to evaluate response. The consensus guidelines of REiNS (Response Evaluation in Neurofibromatosis and Schwannomatosis) state that volumetric analysis of MRI should be used to assess tumor response in NF1-related PN clinical trials.²⁰ Additionally, the phase II placebo-controlled trial of R11577 has provided an historical control group for single-arm trials of new agents in progressive PNs using TTP as the primary endpoint. With these tools in hand, and based on the results from the phase I dose-finding study using PI,¹⁷ we carried out a phase II trial of PI for NF1-associated PN.

In the phase I study, 11/16 patients with tumor-related pain at baseline reported improvement on treatment and 13/14 patients with a palpable tumor had a noticeable decrease in size of the visible component of the tumor, including 4 of 5 patients with proptosis who had a visible improvement in the ability to open the eye. Stratum 2 was designed to more objectively evaluate clinical improvement by restricting eligibility to those patients with specific symptomatology that could be more formally documented. Imaging response was also more consistently evaluated by requiring that all patients undergo MRI scanning sufficient to allow volumetric analyses, both at baseline and throughout the study. Of 17 patients for whom serial 3D MRI analysis was available on the phase I trial, 5 had a 15%–22% volume decrease, only 1 (5.9%) of whom met the criterion for response: ≥20% shrinkage. This is comparable to the results of the current study, where 4 of 82 patients had an imaging response, only 2 of which were confirmed (2.4%), with 6 showing a 15%–17% decrease. Stratum 2 did meet the criteria to exclude a 5% response rate, largely driven by clinical response. However, as has been noted in previous studies with sirolimus,^21,22 clinical response did not correlate with radiographic assessment; none of the 3 patients who met the criteria for clinical improvement had more than 5% shrinkage by volumetric analysis and 1 patient experienced significant worsening or pain despite a 10% shrinkage in tumor volume.

Enrollment on stratum 3 was limited to participants with documented imaging progression prior to enrollment and the same eligibility criteria was used as the placebo arm of the NCI trial with tipifarnib.⁹ Restricting eligibility to those patients with evidence of progression on imaging is critical given that spontaneous cessation of tumor growth can occur. Unlike the results with either tipifarnib⁹ or pirfenidone,⁸ where there was no impact on TTP, or with sirolimus,²³ where there was only a 3.5-month prolongation in TTP, PI resulted in a more than doubling of TTP compared with placebo. Age, which is inversely related to PN growth rate,²⁴ can be a confounding variable when evaluating outcome. However, the median age of patients enrolled on stratum 3 was not significantly different from the placebo arm of the tipifarnib trial, making the significance of the comparison that much more robust.

If response had been used as the primary endpoint, the significant impact of PI would have been minimized or missed, as only 1 patient (3.4%) on stratum 3 had a confirmed imaging response and 3 showed minor volume decreases (range 15%–17%). In comparison, no imaging responses were seen with tipifarnib (n=31), sirolimus (n=46), or pirfenidone (n=36); the largest decrease in tumor volume was 11% on tipifarnib, 12% on the pirfenidone study, and 17% on the sirolimus study.

Many of these tumors become inactive over time, so that at some point, treatment usually becomes unnecessary. Knowledge of this fact may have contributed to the withdrawal of consent by almost 30% of patients on stratum 3 with prolonged SD. Stratum 1, which included patients who were older and asymptomatic, had only one unconfirmed imaging response. Interestingly, the median age was the highest in the subset of patients with pain, with more than half older than 12 years, which may reflect a component of chronic nerve damage. Although subjective improvement in pain was commonly reported on the phase I study, only 2/14 patients (15%) enrolled on the basis of pain in this study met the criteria for clinical improvement.

Interferon-α therapy is associated with a myriad of dose-related side effects, which may be acute and improve over time, such as constitutional symptoms or delayed and chronic side effects such as fatigue, anorexia, and leucopenia. Depressive symptoms in patients receiving PI and ribavirin are common, particularly in those with high baseline depressive scores, which frequently worsen on therapy.²⁵ Behavioral issues, including attention-deficit hyperactivity disorder, are commonly seen in patients with NF1²⁶ and worsening of such and/or fatigue were dose-limiting toxicities in the phase I study.¹⁷ Similarly, treatment-limiting behavioral issues and/or fatigue occurred in 12% and 14% of patients, respectively, in the current study; the majority of behavioral issues were preexisting and worsened on therapy. Four of the 14 patients (28.6%) enrolled in the pain stratum either came off study or required a dose reduction for mood/behavioral issues, most of which were preexisting, and 2 additional patients had dose-limiting fatigue (ie, 43% enrolled on the pain substratum required a dose reduction). The most common nontreatment limiting toxicity (Table 4) was mild (grade 2 and 3) neutropenia, seen in 54/83 (65%), which was not associated with acute infections.

There are no other proven treatment options for patients with actively progressing PN, although emerging targeted therapies have demonstrated promise based on the knowledge of pathways involved in PN growth and progression. For example, recent data from a phase I study of selumetinib, an oral inhibitor of mitogen/extracellular signal-regulated kinase, showed 6/11 patients with progressive PN had an imaging response (≥20% decrease) by volumetric analysis.²⁷ Imaging responses have also been reported with imatinib.²⁸

In conclusion, weekly injections of PI result in at least a doubling of the TTP for patients with progressive PN and is a valid treatment consideration for patients with life-threatening and/or morbid disease during the period of active growth. It also has the advantage of being able to be administered to young children who cannot swallow capsules. Clinical and radiographic improvement can be seen in a subset of patients, although clinical response does not correlate with radiographic changes in symptomatic patients. The optimal duration of therapy is difficult to define given the nuances of the disease, and long-term compliance in this patient population is an issue.

Funding

Merck provided the pegylated interferon alfa-2bas well as support for data management.

Conflict of interest statement. Dr Jakacki was a consultant for AstraZeneca prior to joining the company.

Acknowledgments

Portions of the results of this study were previously presented at the International Society of Pediatric Neuro-Oncology meeting in Toronto, Canada in June 2012.

References

Author notes