Enrollment demographics are shown in . Thirty nine patients were enrolled at Yale University and University of Minnesota between February 2007 and June 2009, of whom 36 were evaluable for response and toxicity. Three were excluded from toxicity assessments for early withdrawal (<1 cycle) for non-compliance (withdrawal from study due to voluntary discontinuation of therapy without protocol-specified toxicities or disease progression). These patients withdrew for personal and social reasons. All drug activity analyses are based on the 39 enrolled patients. The mean age was 64 years, 67% male, 44% and 56% had an ECOG performance status of 0 and 1, and 28%, 39%, and 33% had M1a, M1b, and M1c disease, respectively. Five, four and three patients had acral-lentiginous, ocular or mucosal primaries.
Toxicity assessments and AEs are based on the 36 evaluable patients. The first 17 patients began at 100mg PO BID. Twelve (71%) had drug held for toxicity. The most common adverse events (AEs) requiring cessation of treatment were dyspnea (7 patients, 41%), pleural effusions (6 patients, 35%), fatigue (11 patients, 65%), and diarrhea (2 patients, 12%). The median time on 100mg PO BID was 4 weeks.
Nineteen patients were treated at a starting dose of 70mg PO BID. Dasatinib was held and/or dose-reduced due to toxicity in 9 (47%) of patients. AEs resulting in holding treatment were similar to the 100mg PO BID dose level and included dyspnea (5 patients, 26%), pleural effusion (3 patients, 16%), fatigue (6 patients, 32%) and anorexia (2 patients, 11%). The median time on 70 mg BID dose was 7.5 weeks.
Drug related AEs
AEs are summarized in . The most common AEs were fatigue (35 patients, 97%), dyspnea (31 patients, 86%), pleural effusion (17 patients, 47%), nausea (29 patients, 81%) and anorexia (26 patients, 72%). Most AEs were grade 1 or 2. The most common grade 3 or 4 events were fatigue, dyspnea and pleural effusion. Pleural effusions were commonly successfully managed with a short course of prednisone and/or furosemide.
Drug related clinical adverse events
There were no grade 3 or 4 laboratory AEs. Grade 1–2 events included hypocalcemia (12 patients, 45%), elevations in transaminases (9 patients, 27%), creatinine (8 patients, 23%) and alkaline phosphatase (3 patients, 8%). Hematologic toxicities included neutropenia (3 patients, 8%), thrombocytopenia (4 patients, 11%) and anemia (12 patients, 34%). With the exception of anemia-related fatigue, laboratory abnormalities were not associated with symptoms and resolved with cessation of therapy.
lists all patients demonstrating tumor regression for >16 weeks, their primary site of disease, sites of metastatic disease, and PFS. Of the 36 patients evaluable for activity, two had a confirmed partial response (PR), lasting 64 and 24 weeks, for a RR of 5% (95% CI 1.537–18.145). Tumor regression that did not meet PR criteria was seen in 3 patients (8%), lasting 136, 64 and 28 weeks. The patient with a minor response lasting 64 weeks had 28% reduction in tumor burden. One patient with 27% reduction in tumor burden withdrew from the study after completing the fourth cycle. Regression of liver lesions was observed after 2 cycles in one other patient (not listed in ), but the disease progressed after the fourth cycle. Overall, sites of response included subcutaneous tissues, lymph nodes, lung, liver and skin.
Summary of patients with tumor regression
Progression-free and overall survival curves are shown in . Median PFS was 8 weeks; range 3–136 weeks. Median overall survival was 55 weeks; range 7–159 weeks. As of May 2010, six patients (16%) remain alive, and three (8%) are lost to follow-up.
Progression-free (Panel A) and overall (Panel B) survival in patients treated with dasatinib.
C-kit expression and mutations
Dasatinib is a known potent inhibitor of c-kit, and by the time this study was activated, reports were published documenting excellent clinical responses in melanoma patients whose tumors had activating c-kit mutations treated with c-kit inhibitors 5, 25
. Therefore, in patients demonstrating a tumor response of any kind to dasatinib, and in selected other patients in whom c-kit overexpression and/or mutation was expected based on prior literature reports, we determined c-kit mutation and/or expression by immunohistochemistry (IHC) in pre-treatment tumor samples. Only one responder, a patient with a primary subungual melanoma, had a K642E substitution mutation in exon 13 of c-kit. The other patients demonstrating some tumor reduction all had wild-type c-kit ().
Tissue was available on 5 of 8 other patients with primary acral-lentiginous or mucosal melanomas. An exon 11 deletion mutation was found in one other patient who did not respond to dasatinib. One patient had strong c-kit expression by IHC (3+ on a scale of 0–3), but no mutation, and did not respond. No other responses were observed among this group. Tissue was available for IHC (but insufficient for mutational analyses) on three patients with primary ocular melanoma. All had expression by IHC, none responded.
Dasatinib is an inhibitor of several tyrosine kinases that play a role in melanoma pathogenesis, including src family members, and the EPHA2, PDGFβ and c-kit receptors. We conducted a phase II clinical trial of dasatinib in 39 unselected patients with advanced melanoma.
One of the primary endpoints was to assess toxicity. The starting dose was based on the maximum tolerated dose in the phase I solid tumor trial. Toxicity was intolerable, and a surprisingly high rate of toxicity was seen at lower doses as well. While the majority of adverse AEs were grade 1/2, and all resolved with holding therapy, the chronicity of the toxicities resulted in impaired quality of life and frequent cessation of therapy. Dyspnea and fatigue secondary to pleural effusions were fairly easily managed with temporary treatment cessation and a short course of prednisone, with/without furosemide. The increased frequency and severity of AEs in our study compared with the phase I study likely reflects the improved life expectancy of the patients in this phase II trial and the longer resultant duration of treatment.
The objective RR from dasatinib was 5%. Evidence of tumor regression, confirmed after 4 cycles of therapy, was observed in 5 other patients, including two with PFS exceeding one year. Median PFS was only eight weeks. Pre-clinical studies published after initiation of this trial showed variable inhibition of melanoma cell growth by dasatinib in vitro
. Eustace et al reported an IC50
in the nanomolar range for only 1/5 cell lines 26
, Homsi et al showed variable sensitivity in three cell lines 27
, Buettner et all showed little to no effect on viability 28
, and Woodman et al demonstrated activity in c-KIT mutant cell lines 25
. In our own studies, 2/8 cell lines were growth-inhibited by concentrations <300nM, while the other six were significantly more resistant (Jilaveanu et al, submitted). Peak serum concentrations reached just over 200ηM in patients on the highest dose level in the phase I solid tumor trial (120mg), and the elimination half-life was approximately 4 hours 8, 12
. Thus, although serum drug concentrations might not accurately reflect intracellular levels, the pharmacokinetic data raise concerns that levels of drug necessary for cell growth inhibition may not be achievable in most patients. Of note, in the phase I studies, clinically tolerable doses were sufficient to inhibit c-Src in post-treatment tumor biopsy samples. However, inhibition of Src-kinases may not be sufficient to mediate a measurable anti-tumor effect in most patients. The phase II clinical study was not designed to detect other potentially meaningful biological effects of src inhibition, such as reduction in motility or invasion 28
After this trial was written, reports of c-kit mutations in up to 30% of acral-lentiginous melanomas and 39% of mucosal melanomas were published 29
. Over the past 4 years, several investigators have reported dramatic tumor regression in melanoma patients whose tumors had c-kit mutations, as seen in GIST patients. In contrast, c-kit mutations were generally not found in cutaneous melanomas arising in intermittently sun-exposed skin or in ocular melanomas, although the latter often demonstrated c-kit overexpression by IHC 30, 31
. Two of our 9 patients with primary acral-lentiginous or mucosal primaries had documented c-kit mutations. Both had subungual primaries, one achieved a partial response to dasatinib. No confirmed responses were observed in c-kit overexpressing tumors lacking a c-kit mutation, although we did not assess gene amplification status. Our data are consistent with other reports 5, 25
. Upon disease progression, the responding patient with a tumor c-kit mutation developed an excellent response to sorafenib, another c-kit inhibitor, in combination with temozolomide. The systemic response was durable, but she ultimately developed uncontrollable brain metastases. This limited single patient experience suggests that different c-kit mutations may respond variably to c-kit inhibitors, that progression on one c-kit inhibitor does not preclude response to another, and lack of drug CNS penetration and consequent CNS disease progression may be a problem.
Other than the patient with a subungal melanoma and a c-kit mutation, other patients who had tumor reduction had tumors that were wild type for c-kit, indicating that other pathways and molecules targeted by dasatinib might be driving proliferation in these tumors. In our in vitro
studies, we attempted to identify predictors of sensitivity to dasatinib, and found that high caveolin-1 levels are associated with in vitro
growth inhibition (Jilaveanu et al, submitted). Although the number of patients in whom activity was seen in this trial is small, there might be an association between pre-treatment caveolin-1 tumor levels and tumor reduction. This is consistent with findings in other diseases 32, 33
, and requires validation in additional trials of dasatinib for melanoma. The biological relationship between dasatinib target inhibition, caveolin-1 expression, and the mechanism of growth inhibition remains unclear.
Based on the study design, an overall response rate ≥20% or 6-month PFS ≥20%, single agent dasatinib would have been considered worthy of further study in this disease. This is a relatively stringent requirement for a multi-target kinase inhibitor without preselection of patients based on tumor biology. Our results do suggest that dasatinib has some activity in melanoma, and further evaluation of patient subsets such as caveolin-1 over-expressers and c-kit mutated tumors are ongoing in the preclinical and clinical setting, respectively. Further evaluation of dasatinib in carefully selected melanoma patients might be warranted.
As with other anti-neoplastic agents, increased activity might be seen with addition of drugs that work by alternative mechanisms. Combinations with other small molecule inhibitors might result in severe toxicities, as has been the experience when combining small molecule inhibitors in other diseases 34
. Addition of cytotoxic chemotherapy with limited overlapping toxicities might be better tolerated. Our recent in vitro
studies, and those of others, suggest that addition of cisplatin or dacarbazine might enhance dasatinib activity 27
. A trial of a chemotherapy-dasatinib combination has been initiated at another center. It may be important to identify those patients that have the best chance for responding to the combination, for example, by pre-selecting for high caveolin-1 expression in tumor, and for factors permissive of response to the chemotherapy, such as low MGMT expression and an intact DNA mismatch repair pathway.
In summary, single agent dasatinib has minimal activity in unselected melanoma patients, and at doses used in this trial, produces poorly tolerated toxicities in most patients. Tumor regression was noted in approximately 14% of patients without tumor c-kit mutations and one of two with c-kit mutations. This level of activity suggests that biomarker-based patient pre-selection may identify a subset of patients who could potentially derive benefit from dasatinib as a single agent or in combination with other agents, and predictors of sensitivity/response need to be studied further. Because of the AEs, future combination studies should be based on biological activity achievable at lower serum concentrations (perhaps sufficient to inhibit src) and molecularly defined patient subsets.