Phase II Study of Nilotinib in Melanoma Harboring KIT Alterations Following Progression to Prior KIT Inhibition

Alterations in the KIT proto-oncogene define one unique molecular subset of melanoma. Mutations and amplification of KIT are observed in 3% of all melanomas, and are more common in disease arising from mucosal, acral, or chronically sun-damaged surfaces (1). The mutations identified are, in most cases, substitution mutations mutually exclusive of BRAF and NRAS mutations and often affect the juxtamembrane or kinase domains of KIT, leading to constitutive activation of KIT tyrosine kinase activity.

The clinical activity of KIT inhibition in those melanomas driven by KIT alterations has been reported in patients treated with agents such as imatinib (2–4), dasatinib (5), sorafenib (6), and sunitinib (7), with efficacy observed in prospective trials of imatinib (8–10) and sunitinib (11). Despite the clinical benefit achieved with KIT inhibition in select patients with melanoma harboring KIT mutations, most patients ultimately experience disease progression. Failure of these agents has been observed within the brain (12), which may be related to the frequent development of brain metastases in patients with advanced melanoma, as well as the limited central nervous system (CNS) penetration of many small-molecule kinase inhibitors.

Secondary resistance to KIT inhibition in patients with gastrointestinal stromal tumors (GIST), a disease characterized by activating deletions or insertions in KIT, is caused primarily by the development of secondary KIT mutations commonly affecting the tyrosine kinase domains (13). There can additionally be outgrowth of resistant subclones present at baseline that are selected during KIT inhibitor therapy. In GIST, the use of alternative KIT tyrosine kinase inhibitors after progression on imatinib, including sunitinib (14), sorafenib (15), and regorafenib (16), has proven beneficial; however, the efficacy of sequential KIT inhibitors in melanoma is unknown.

Nilotinib (Tasigna, AMN107) is a tyrosine kinase inhibitor structurally derived from imatinib that is approved in the United States for the treatment of chronic and accelerated phase Philadelphia chromosome-positive chronic myelogenous leukemia in patients resistant or intolerant to prior therapy with imatinib. Nilotinib binds to and inhibits the kinase domain of ABL/BCR-ABL and of the DDR, KIT, PDGF, and several EPH receptor kinases with greater potency than imatinib (17, 18), and maintains activity against a range of exon 9, 11, and 13 KIT mutations (19). We conducted a phase II trial of nilotinib in patients with melanoma harboring KIT aberrations who experienced disease progression or intolerance to a prior KIT inhibitor. Given the frequent complication of brain metastases in patients with this disease and the potential for second-generation inhibitors of KIT to have activity within the CNS (20), a cohort of patients with brain metastases was included.

The primary objective was to assess the efficacy of nilotinib in patients with metastatic melanoma arising from acral, mucosal, or chronically sun-damaged surfaces characterized by mutations or amplification of KIT after demonstration of disease progression or intolerance to a prior KIT tyrosine kinase inhibitor. Secondary objectives included efficacy assessment of nilotinib in patients with advanced KIT-mutant melanoma and CNS metastases. Tumor samples from all patients were prospectively tested for KIT mutation or amplification by qPCR) or FISH as previously described (8, 10).

Patients who met eligibility criteria received nilotinib 400 mg by mouth twice daily. Safety evaluations, including clinical and laboratory assessments, were conducted at baseline, every week for 4 weeks, every 2 weeks for 4 weeks, every 4 weeks for 28 weeks, and then every 3 months subsequently. Adverse event severity was graded using the NCI Common Terminology Criteria for Adverse Events, v3.0. Tumor response was measured radiographically every 8 weeks for 32 weeks and every 12 weeks subsequently using RECIST 1.0 criteria, and included brain imaging for those with CNS involvement. Patients remained on study until the time of progression or the development of unacceptable toxicity not manageable with dose modification.

The primary endpoint was the proportion of patients who were alive and without progression of disease 4 months after beginning treatment with nilotinib. Secondary endpoints included best overall response rate (BORR), time-to-progression (TTP), overall survival (OS), and tolerability.

Patients were enrolled from eight academic medical centers between January 23, 2009 and June 14, 2011. Eligible patients had advanced melanoma harboring a KIT mutation or amplification and arising from acral, mucosal, or chronically sun-damaged surfaces, as documented by the presence of solar elastosis. Patients without CNS metastases were enrolled onto cohort A and must have experienced disease progression or intolerance to one or more KIT tyrosine kinase inhibitors. Intolerance was defined as drug discontinuation due to grade-2 events persisting for one month or longer, or any grade-3 or grade-4 rash, fluid retention, cardiopulmonary events, thrombocytopenia, liver function abnormalities, or diarrhea that persisted despite optimal supportive care measures. Patients with measureable CNS disease harboring a KIT mutation were enrolled onto cohort B and did not require prior therapy for eligibility. For those who received prior radiotherapy for CNS disease, progression was required in previously treated lesions or new lesions must have developed.

Other key inclusion criteria included age greater than 18 years, life expectancy greater than 3 months, Eastern Cooperative Oncology Group performance status of zero, one, or two, measurable disease by Response Evaluation Criteria in Solid Tumors (RECIST) v1.0, and adequate organ function. Exclusion criteria included prior therapy with nilotinib and clinically significant heart disease. All patients provided written-informed consent before initiating study procedures. The study was reviewed and approved by Institutional Review Boards at all participating centers.

Cohort A used an optimal Simon, two-stage design with 87% power to compare a null disease control rate (DCR) of 5% with an alternative of 25%, with a one-sided type-I error of 7.5%. The target sample size was 28 patients, of whom 25 were expected to be evaluable for outcome. In the first stage, 13 evaluable patients would be assessed. If 2 or more patients achieved 4-month disease control, an additional 12 evaluable patients would be assessed in the second stage. If 3 or more of 25 achieved 4-month disease control, then nilotinib would be considered promising in this disease setting. A second feasibility cohort of 10 patients (cohort B) was added after the study began to estimate the 4-month DCR in patients with advanced, KIT-mutated melanoma and CNS metastases. Nilotinib would be of interest in this cohort if at least 2 of 10 patients achieved 4-month disease control.

Baseline patient characteristics and adverse events were summarized using descriptive methods. Adverse events were reported as the most severe manifestation of each event category during any cycle of treatment. Four-month DCR was defined as the proportion of treated patients with a complete or partial response (PR), or stable disease (SD) per RECIST 1.0 after 4 months of therapy. BORR was defined as the proportion of treated patients with either complete or PR (per RECIST) as best response to therapy. The number of treated patients in each cohort was the denominator for estimates of DCR and BORR. TTP was defined as the time from initiation of nilotinib to the date of progression or last follow-up. OS was defined as the time from initiation of nilotinib to the date of death or last follow-up. Four-month DCR and BORR are presented with 90% exact binomial confidence intervals. TTP and OS are presented using the method of Kaplan–Meier, with point-wise 90% confidence intervals (CI) estimated using log(-log(survival)) methodology.

Dr. Hodi developed the original study design and was responsible for the IND. Novartis provided investigational drug in addition to funding, and was involved in study design that was developed in conjunction with the authors. The study sponsor had no role in the data collection, the data analysis, data interpretation, writing of the report, or the decision to submit for publication.

Twenty patients were enrolled (11 in cohort A and 9 in cohort B) and 19 treated on this study (11 in cohort A and 8 in cohort B). One patient who enrolled in cohort B withdrew consent before receiving study therapy. With the completion of a series of studies of imatinib and other agents targeting KIT in patients with melanoma harboring KIT alterations, enrollment to second-line trials became increasingly challenging and enrollment to this trial was closed before completion of either the first stage of the two-stage trial or the CNS feasibility component. The design for cohort A was modified to a single-stage design with 11 patients, with 87% power to compare a null DCR of 5% with an alternative of 39.5%, using an exact binomial test and a one-sided type-I error of 7.5%.

Baseline patient characteristics of the 19 treated patients are shown in Table 1. Patients were predominantly female (74%), with a median age of 67 years (range, 38–85 years). Twelve (63%) patients had mucosal melanoma, 4 (21%) had acral melanoma, and 3 (16%) had melanoma arising from chronically sun-damaged skin (CSD). All patients had locoregionally advanced (5%) or distant disease (95%); most patients received one or more prior therapies. Sixteen patients received prior imatinib, one received both sorafenib and imatinib (patient 2), and 3 received prior ipilimumab therapy. All patients previously treated with a KIT inhibitor experienced progression on those agents and were not enrolled onto this study due to intolerance of prior therapy. Six of the 8 patients treated on cohort B received prior therapy with imatinib, and 2 patients were naïve to KIT inhibition. Patient demographic and disease characteristics were similar between cohorts A and B.

Tumor from the 19 treated patients was tested for the presence of KIT mutations, with 17 harboring such alterations (Tables 2 and 3). The specific mutations identified included exon 11 L576P (n = 4), exon 11 V560D (n = 1), exon 11 V560E (n = 1), exon 11 W557R (n = 1), exon 11 V559C (n = 1), exon 11 WKVVE 557-561 (n = 1), exon 13 K642E (n = 3), exon 13 Y646D (n = 1), exon 17 D820Y (n = 1), exon 17 N822K (n = 1), and exon 18 L831P (n = 1). One patient had tumor harboring two exon 13 mutations (R634Q and K642E). KIT amplification was tested in 12 cases, with 8 found to harbor such alteration. Two cases harbored amplification without a concurrent KIT mutation.

At the time of data analysis, 18 of the 19 treated patients were off-study, 14 of whom due to progressive disease. Median follow-up was 16.2 months in cohort A (90% CI, 6.9–37.5 months) and 11.7 months in cohort B (90% CI, 2.1 months–∞).

Adverse events classified as possibly, probably, or definitely related to nilotinib are shown in Supplementary Table S1. Events that were recorded multiple times for any patient are reported only once according to the worst grade. Although nilotinib was generally well tolerated, 17 of the 19 patients treated reported adverse events, with fatigue (26%) and low-grade musculoskeletal and gastrointestinal discomfort (32%) most commonly observed. Grade-3 toxicities were observed in 4 patients, and included rash (n = 1), elevated pancreatic enzymes (n = 2), and transaminitis and hyponatremia (n = 1). Grade 3 toxicity was managed by dose reduction to 400 mg QD (n = 2) or dose delay followed by reinitiation of treatment at 400 mg twice a day (n = 2). No patient experienced grade-4 related adverse events. Toxicity rates and patterns were comparable for cohorts A and B.

Four-month disease control rate. In cohort A, 3 of 11 patients were alive without disease progression at 4 months (27%; 90% CI, 8%–56%), a proportion significantly greater than the DCR of 5% (P = 0.03) assumed under the null hypothesis. On the basis of three observed responses, there is sufficient evidence to conclude that nilotinib would have been considered worthy of further study in cohort A based on the initial two-stage design. In cohort B, 1 of 8 treated patients achieved disease control at 4 months (12.5%; 90% CI, 0.6%–47%), with no evidence that 4-month DFR is greater than 5% in this population.

Response rate. Of the 19 patients treated, 4 were inevaluable for radiographic response to therapy in non-CNS lesions. In cohort A, patient 11 initiated therapy but subsequently underwent resection of abdominal disease due to tumor-associated gastrointestinal bleeding. In cohort B, patient 16 initiated therapy but developed rapid clinical decline due to progressive leptomeningeal disease and withdrew consent for further treatment and evaluation. Patients 15 and 17 presented with CNS-only disease, without measurable lesions in extracranial sites.

In cohort A, two PRs were observed (18.2%, 90% CI, 3%–47%). One PR was observed in an 81-year-old female with advanced vulvar melanoma harboring an exon 11 L576P mutation without concurrent amplification (Patient 3). She previously achieved a durable PR to therapy with imatinib lasting 12.4 months and has an ongoing response to nilotinib at 37.5 months. Additional patients achieved minor responses to therapy (Tables 2 and 3).

Of the 8 patients treated on cohort B, 7 were evaluable for response in CNS metastases which either were not previously treated with radiotherapy or which demonstrated progression following treatment (Fig. 1). Assessing CNS lesions only, we observed one PR (12.5%, 90% CI, 0.6%–47%) lasting 3.9 months (Patient 4) and one minor response (Patient 15), each in patients not previously treated with a KIT inhibitor. The PR was observed in a 48-year-old female with mucosal melanoma arising from the anorectal region harboring an exon 11 V560D mutation without concurrent amplification. A brain MRI performed 5 months after receiving stereotactic radiosurgery to left temporal, left parietal, right frontal, and right mid-cerebellar lesions demonstrated the development of progression in the previously treated lesions and the development of numerous infra and supratentorial hemorrahgic brain metastases (Fig. 1A). She achieved a minor response in her extracranial metastases (20% tumor regression by RECIST criteria) and a PR in her target brain metastases (36% regression by RECIST criteria) as demonstrated by the circled lesions in Fig. 1A and 1B. Despite durable stability in her extracranial disease after 4 months of therapy and further reduction in the size of several of the brain metastases, there was progression in nontarget brain metastases and she was taken off study.

Time-to-progression. The TTP achieved with nilotinib as well as to a prior KIT inhibitor, if applicable, is shown by patient in Fig. 2. The median TTP was 3.4 months (90% CI, 0.9–5.5 months) and 2.6 months (90% CI, 1.8–3.9 months; Fig. 3A) in cohorts A and B, respectively.

Overall survival. Eleven patients (57.9%) were deceased at the time of data analysis, with one patient lost to follow-up. The median OS in cohort A was 14.2 months (90% CI, 7.1 months–∞) and was longer than observed in cohort B (4.3 months; 90% CI, 3.5–11.9 months; P = 0.05; Fig. 3B).

These results demonstrate that a subset of patients with melanomas harboring genetic alterations of KIT may benefit from nilotinib after experiencing disease progression to a prior KIT inhibitor. Three of 11 patients without brain metastasis achieved disease control at 4 months with nilotinib, with observed progression-free survival times of 5.5, 11.5, and 37.5+ months. Notably, patients 3 and 20 achieved a durable PR and CR, respectively, to imatinib lasting 12.4 and 20 months, respectively, before achieving durable PRs to nilotinib, demonstrating that nilotinib can overcome the development of secondary resistance to imatinib. On the basis of the original study design for cohort A which required 3 or more patients to achieve disease control at 4 months, the primary endpoint of 4-month DCR was achieved.

Given the high incidence of brain metastases in melanoma and the potential efficacy of second-generation KIT inhibitors in CNS metastases (20), we included an exploratory cohort of patients with brain metastases from melanoma harboring KIT alterations. Although available data suggest the limited penetration of nilotinib within the CNS, clinical activity has been observed in the brain in BCR-ABL–positive leukemia (21). Such efficacy may be explained by the high protein-binding affinity of nilotinib coupled with the low protein concentration within the cerebrospinal fluid, thus resulting in relatively higher amounts of free nilotinib within the CNS. Indeed, of 7 patients in our trial evaluable for response in brain lesions, one achieved a 36% reduction and another achieved a 25% reduction in the CNS tumor burden with therapy. A mixed response in the brain lesions was observed in some cases, with clear reduction in the size of several brain metastases and unambiguous progression in others. Although anecdotal, these variable responses may suggest more prominent intra-tumoral molecular heterogeneity in CNS lesions when compared with disease in other organs or variable pharmacologic penetration into the brain metastases. Of note, both patients who achieved radiographic responses within the brain were not previously treated with a KIT inhibitor such as imatinib. Despite the radiographic changes observed, the progression-free and OS in this cohort of patients were short.

The greater potency of nilotinib over imatinib against the mutant KIT oncoprotein provides pharmacologic rationale for using nilotinib (18, 22). Furthermore, the sensitivity of specific KIT mutations to clinically available inhibitors can differ, with some mutations affecting the binding affinity of specific inhibitors of KIT as previously demonstrated in in vitro and clinical studies of GIST (13, 19, 23). Although preliminary evidence of activity with nilotinib in patients with melanoma harboring KIT alterations not previously treated with a KIT inhibitor has been observed, with two PRs lasting 8.4 and 10+ months reported in nine patients with melanoma harboring a KIT alteration not previously treated with a KIT inhibitor (24), whether nilotinib is superior to imatinib in KIT-inhibitor naïve patients with melanoma is unknown. In advanced GIST, nilotinib was not superior to imatinib as first-line therapy and did not improve outcomes when compared with best-supportive care in the third-line setting (25, 26). Importantly, mechanisms of secondary resistance in GIST, which commonly involve the development of secondary KIT mutations affecting the tyrosine kinase domains in exons 13 and 17 (27,28), appear to differ from those observed in melanoma driven by KIT alterations. Thus far, no such secondary mutations have been observed in KIT melanoma. Rather, the limited data available suggest that, in melanoma, the development of secondary NRAS mutations (11) and activation of the mTOR pathway by alternative mechanisms may result in secondary resistance (29).

In conclusion, the use of nilotinib in a subset of patients with melanoma harboring KIT alterations previously treated with an inhibitor of KIT can result in clinical benefit, although efficacy of this agent in brain metastasis is limited. Although this trial is underpowered to conclude clinical benefit, the data suggest further studies of sequential KIT inhibitor therapy for this molecular subset of patients is warranted.

R.D. Carvajal is a consultant/advisory board member for AstraZeneca, Aura Biosciences, and Novartis. J. Weber is a consultant/advisory board member for Bristol-Myers Squibb, GlaxoSmithKline, Merck, and Novartis. P.B. Chapman is a consultant/advisory board member for Bristol-Myers Squibb, GlaxoSmithKline, and Roche/Genentech. M.C. Heinrich is a consultant/advisory board member for and reports receiving commercial research grants and speakers bureau honoraria from Novartis; has ownership interest (including patents) in MolecularMD; and has provided expert testimony in patent litigation proceedings related to a Korean patent concerning treating GI stromal tumors with imatinib for Novartis. B.C. Bastian is a consultant/advisory board member for Novartis. C.L. Corless reports receiving speakers bureau honoraria from and is a consultant/advisory board member for Novartis. F.S. Hodi reports receiving other commercial research support from and is a consultant/advisory board member for Novartis. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.J. O'Day, J.D. Wolchok, A. Giobbie-Hurder, M.C. Heinrich, F.S. Hodi

Development of methodology: D.P. Lawrence, S.J. O'Day, A. Giobbie-Hurder, M.C. Heinrich, J.A. Fletcher, F.S. Hodi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.D. Carvajal, D.P. Lawrence, J. Weber, T.F. Gajewski, R. Gonzalez, J. Lutzky, S.J. O'Day, O. Hamid, J.D. Wolchok, P.B. Chapman, R.J. Sullivan, J.B. Teitcher, C.R. Antonescu, M.C. Heinrich, C.L. Corless, J.A. Fletcher, F.S. Hodi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.D. Carvajal, D.P. Lawrence, J. Weber, T.F. Gajewski, S.J. O'Day, O. Hamid, N.H. Ramaiya, A. Giobbie-Hurder, C.R. Antonescu, M.C. Heinrich, C.L. Corless, J.A. Fletcher, F.S. Hodi

Writing, review, and/or revision of the manuscript: R.D. Carvajal, D.P. Lawrence, J. Weber, T.F. Gajewski, R. Gonzalez, J. Lutzky, S.J. O'Day, O. Hamid, J.D. Wolchok, P.B. Chapman, R.J. Sullivan, J.B. Teitcher, N.H. Ramaiya, A. Giobbie-Hurder, C.R. Antonescu, M.C. Heinrich, B.C. Bastian, J.A. Fletcher, F.S. Hodi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R.D. Carvajal, D.P. Lawrence, C.L. Corless, F.S. Hodi

Study supervision: R.D. Carvajal, D.P. Lawrence, R. Gonzalez, O. Hamid, J.D. Wolchok, F.S. Hodi

The authors acknowledge the assistance of Carla Becerra, Maria Gargano, and Janice Russell from Dana-Farber Cancer Center and Shahnaz Singh-Kandah from Memorial Sloan Kettering Cancer Center for their contributions to this study, which included data acquisition, data analysis, and administrative support.

Funding for this trial was provided by Novartis. R.D. Carvajal and B.C. Bastian received funding from NIH/NCI grant P01 CA025874.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.