Safety and efficacy of arsenic trioxide for patients with advanced metastatic melanoma


  • Ahmad A. Tarhini MD, MSc,

    1. Department of Medicine, Division of Hematology/Oncology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    2. Melanoma and Skin Cancer Program, University of Pittsburgh Cancer Institute, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • John M. Kirkwood MD,

    1. Melanoma and Skin Cancer Program, University of Pittsburgh Cancer Institute, Pittsburgh, Pennsylvania
    2. Department of Medicine, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Hussein Tawbi MD, MSc,

    1. Department of Medicine, Division of Hematology/Oncology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    2. Melanoma and Skin Cancer Program, University of Pittsburgh Cancer Institute, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • William E. Gooding MS,

    1. Biostatistics Facility, University of Pittsburgh Cancer Institute, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Mohammed F. Islam MD,

    1. Medical Center Clinic, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Sanjiv S. Agarwala MD

    Corresponding author
    1. Medical Oncology, St. Luke's Health System, Bethlehem, Pennsylvania
    • Department of Medical Oncology, St. Luke's Health System, 801 Ostrum Street, Bethlehem, PA 18015
    Search for more papers by this author
    • Fax: (610) 954-2108


This article is corrected by:

  1. Errata: Erratum Volume 119, Issue 4, 924, Article first published online: 28 August 2012



Arsenic trioxide (ATO) cytotoxicity and apoptosis induction has been demonstrated with numerous cancer cell lines, including human melanoma.


A second-line, phase 2, single-arm study of ATO was conducted in patients with inoperable American Joint Committee on Cancer (AJCC) stage IV melanoma. One cycle consisted of a loading dose of 0.32 mg/kg/day for 4 days in Week 1, followed by 0.25 mg/kg/day twice per week for 6 weeks, followed by 1 week of rest, at which time response assessment was performed.


Twenty-one patients (median age, 63.8 years) were accrued. All had stage IV melanoma including M1a (2 patients), M1b (6 patients), and M1c (13 patients) disease. One patient had metastatic choroidal melanoma and 20 patients had cutaneous melanoma. Twenty patients had received prior therapy. Possible treatment-related grade 3 of 4 toxicities (using the National Cancer Institute Common Toxicity Criteria) included 1 case of idiopathic thrombocytopenic purpura and 1 case of elevated lactate dehydrogenase. Four patients did not complete the first cycle of therapy and were not evaluable for response. Among 17 evaluable patients, 1 patient (6%; 95% confidence interval [95% CI], 0–29%) achieved a partial response lasting 7 months, and 10 patients (59%) had disease stabilization after at least 1 cycle, but all eventually developed disease progression. The median time to disease progression was 17 weeks (95% CI, 11–38 weeks) and the median survival was 13 months (95% CI, 12–26 months).


ATO as tested in the current trial was found to be well tolerated and had limited activity in patients with metastatic melanoma. The application of this agent in combination with either chemotherapy or agents that target recognized critical signaling and antiapoptotic pathways of melanoma has not yet been performed. Cancer 2008. © 2008 American Cancer Society.

Advanced metastatic melanoma is refractory to most standard systemic therapy, with a median survival of 6 to 12 months. To our knowledge, dacarbazine (DTIC) remains the only chemotherapeutic agent approved by the U.S. Food and Drug Administration (FDA) since 1976 with short-lived response rates not exceeding 10% to 20%.1 The addition of other chemotherapeutic agents to DTIC, such as cisplatin, nitrosoureas, and vinca alkaloids has been reported to yield higher response rates, up to 40%, but has failed to impact patient survival. Most notable is the 4-drug combination of cisplatin, carmustine, DTIC, and tamoxifen (CBDT) or the “Dartmouth regimen,” which yielded high response rates (up to 55%), but a randomized phase 3 trial of CBDT versus DTIC alone demonstrated no significant difference in survival between the 2 groups.2

To our knowledge, biologic agents including interleukin-2 (IL-2) and interferon‒α (IFN) have produced the most promising results to date in melanoma. Recombinant IL-2, a cytokine with a range of immunomodulatory effects investigated in high-dose regimens in prospective clinical trials, demonstrated a response rate of approximately 16%.3 High-dose (HD) IL-2 was approved by the U.S. FDA based on a retrospective analysis of 270 patients with metastatic melanoma who were treated with HD IL-2 between 1985 and 1993 and who demonstrated objective response rates of 16%, including 6% complete responses. The median duration of response was 6.5 months and 60% of complete responders remained progression-free at 5 years.4 The major toxicity and cost associated with HD IL-2 has restricted its use to selected patients at major medical centers. HD IFN also received U.S. FDA approval in the adjuvant setting after demonstrating a recurrence-free and overall survival benefit in randomized clinical trials.5–7

Combinations of chemotherapy and biologic agents (biochemotherapy) have been extensively tested in melanoma, but large randomized clinical trials failed to demonstrate any significant benefit in terms of survival or time-to-progression for the biochemotherapy arm compared with chemotherapy alone.8, 9 Clearly, new therapeutic alternatives for metastatic melanoma are a major clinical need.

The use of heavy metals in the treatment of neoplastic disease has a significant precedent, with the antitumor activity of cisplatin a prominent example. Arsenic compounds were known to Chinese and Indian medicine for more than 1000 years. Historically, arsenic has been topically effective for superficial tumors, skin disease, and chronic myeloid leukemia, but since the 1960s has not been commonly used in the U.S. Arsenic compounds have been regarded as comutagens and cocarcinogens by epidemiologic studies.10

Arsenic trioxide (ATO) has been shown to be a promising new agent in the treatment of both solid and hematologic tumors. Reports from Shanghai and the Harbin Institute of Hematology in China showed that ATO at a dose of 10 mg/day by intravenous infusion for 28 to 54 days can induce remission in acute promyelocytic leukemia (APL) patients, even for those who had a recurrence after all-trans retinoic acid (ATRA) with or without chemotherapy.11 No significant toxicity was observed including bone marrow suppression. Subsequently, U.S. studies with a new formulation of arsenic trioxide (Trisenox; Cell Therapeutics, Seattle, Wash) demonstrated similar results of ATO in inducing complete remissions in patients with APL who have developed a disease recurrence after ATRA with or without chemotherapy.12, 13 The clinical response was associated with partial cytodifferentiation and the induction of apoptosis through caspase activation in leukemic cells and down-regulation of bcl-2 expression.

ATO has been tested in numerous human cancer cell lines, including human melanoma,14, 15 and apoptosis has been demonstrated in the majority of cell lines. The mechanism(s) of apoptosis were attributed to multiple mechanisms, including bcl-2 down-regulation, caspase 3 activation, tubulin dysfunction, and inhibition of NF-κB.16–21 Postulated mechanisms of action of arsenic-induced tumor cell cytotoxicity, as described in vitro and ex vivo studies, include tubulin dysfunction,22 stimulation of c-Jun NH2-terminal kinase-dependent and p53-independent pathway,23 increasing mitochondrial permeability transition pore,24 inhibition of the MAP kinase cascade, activation of NADPH oxidase and reactive oxygen species,25 caspase activation and down-regulation of bcl-2,20, 26 modulation of the glutathione redox system,27 and inhibition of NF-κB.21, 28

In the phase 1 dose escalation study in patients with advanced hematologic malignancies, 25 doses of ATO, ranging from 0.10 to 0.25 mg/kg/day, were given throughout 4 to 5 weeks. Additional courses were administered beginning 3 to 5 weeks after the preceding course. In the phase 1 study conducted in patients with solid tumors, patients received 5 daily infusions of ATO in doses ranging from 0.15 to 0.35 mg/kg/day. Additional courses were administered at 4-week intervals. Fluid retention, dyspnea, and fatigue were dose-limiting at a daily dose of 0.35 mg/kg for 5 days. Less common reactions included hyperglycemia, skin rash, diarrhea, headache, and prolonged QTc interval (QT corrected for heart rate) on electrocardiogram. Pharmacokinetic data from these and other studies, including the prolonged half-life of ATO, suggested a regimen consisting of a loading phase in Week 1, followed by a maintenance phase in which ATO is administered less frequently than would be optimal.29, 30

Finally, arsenic compounds have been regarded as comutagens and cocarcinogens by epidemiological studies. Carcinogenicity studies have not been conducted with Trisenox by intravenous administration. The active ingredient of Trisenox, arsenic trioxide, is a human carcinogen.



Patients were eligible if they were age ≥18 years and had histologically confirmed metastatic melanoma (American Joint Committee on Cancer [AJCC] stage IV), and had measurable or evaluable disease. All patients were required to sign an informed consent.

Patients may have been previously untreated or may have received ≤1 prior chemotherapy and/or ≤2 biological therapies. At least 4 weeks must have elapsed since prior therapy (6 weeks for nitrosoureas or mitomycin C). Prior radiotherapy was required to be completed at least 4 weeks before study drug administration. The patient must have recovered from all toxicities attributable to prior therapy.

Eligible patients were also required to have met the following criteria: an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2; life expectancy of ≥12 weeks; and adequate hematologic values (absolute neutrophil count ≥1500 cells/μL, hemoglobin ≥9 g/dL, and platelet count ≥100,000/μL), hepatic function (serum bilirubin ≤2 × the upper limit of normal [ULN] and aspartate aminotransferase [AST] and alanine aminotransferase [ALT] ≤2 × ULN or, in the presence of documented liver metastases, AST and ALT ≤5 × ULN), and renal function (serum creatinine ≤1.5 × ULN). Fertile patients were required to use an appropriate method of contraception and a negative pregnancy test was required for women of childbearing potential.

Patients were ineligible if they had any of the following: 1) history of or active brain metastases; 2) history of significant cardiovascular disease; 3) active infection; 4) pregnant or lactating women; 5) concurrent second malignancy except for squamous cell or basal carcinoma of the skin, stage I prostate cancer, or cervical intraepithelial neoplasia (any other malignancy must have been in continuous clinical remission for at least 5 years before study enrollment); and 6) an absolute QTc interval >460 milliseconds (msec) in the presence of serum potassium >4.0 mEq/dL and magnesium values >1.8 mg/dL.

Study Design and Treatment

This was a phase 2, single-arm study performed in an outpatient oncology unit at the University of Pittsburgh Cancer Institute. Eligible patients received a loading dose of ATO (Trisenox) at 0.32 mg/kg/day for 4 consecutive days in Week 1. This was followed by a maintenance dose of 0.25 mg/kg/day twice per week for 6 weeks, followed by a week of rest, for a total cycle length of 8 weeks (Fig. 1). ATO was administered as a 1-hour intravenous infusion and was continued until the occurrence of either unacceptable toxicity or evidence of disease progression. Patients without unacceptable toxicity or disease progression were allowed to continue on study for a maximum of 12 cycles of therapy.

Figure 1.

Treatment cycles. ATO indicates arsenic trioxide; iv, intravenously.

Because of the risk of QTc interval prolongation with ATO and the consequent risk of torsade de pointes, serum electrolytes (potassium and magnesium) and creatinine were assessed before initiating therapy with ATO. Preexisting electrolyte abnormalities were corrected and, if possible, concomitant drugs that deplete electrolytes or prolong the QT interval were avoided or discontinued. A12-lead electrocardiogram was performed to document a baseline QTc interval that was <460 msec.

Toxicity and Response Assessments

The National Cancer Institute's Common Toxicity Criteria (version 3.0) was used for grading toxicities. At each visit, each patient was queried regarding adverse events. Patients were monitored for toxicity initially on Day 1 of each treatment cycle, with a complete history and physical examination, vital signs, electrocardiogram, and a laboratory assessment (including complete blood counts, coagulation profile, serum chemistry including potassium and magnesium, and urinalysis). The same assessment was repeated at least weekly during a treatment cycle and at the end of the cycle.

Systemic computed tomography (CT) scans and gadolinium-enhanced magnetic resonance imaging (Gd-MRI) of the brain were performed at baseline. CT scans (and MRI brain only as clinically indicated) were repeated at the completion of every even-numbered treatment cycle to assess response. Radiologic studies were reviewed and confirmed independently by a central radiology facility, which was blinded to patient characteristics.

The following criteria (modified from World Health Organization [WHO] reporting of response) were used per study protocol to determine the levels of response. A complete response (CR) was defined as the complete disappearance of all clinically detectable malignant disease; a partial response (PR) was defined for bidimensionally measurable disease as a decrease by at least 50% of the sum of the products of the largest perpendicular dimensions of all measurable lesions and, for unidimensionally measurable disease, a decrease by at least 50% in the sum of the largest dimensions; and stable disease (SD) was defined as no change in the size of tumor lesions or an increase or decrease of <25% and the absence of any new lesions. Levels of response were determined separately at 2 consecutive examinations at least 4 weeks apart and were valid only in the absence of the development of new central nervous system lesions.

Dose Modifications

Treatment with ATO was interrupted, adjusted, or discontinued before the scheduled end of therapy at any time toxicity greater than grade 3 on the NCI Common Toxicity Criteria was observed and judged to be possibly related to ATO treatment. Patients were allowed to resume treatment only after resolution of the toxic event or after recovery to baseline status of the abnormality that prompted the interruption. In such cases, treatment resumed at 50% of the preceding daily dose. Patients who experienced a recurrence of toxicity were removed from treatment. If ATO-related grade 2 or 3 toxicities were present at the scheduled start of a treatment cycle, that cycle was allowed to be delayed for up to 2 weeks; patients with symptoms persisting at 4 weeks after the last study treatment were withdrawn from the study.

Statistical Methods

This was a single-center, open-label, single-arm phase 2 study in patients with metastatic melanoma who had received prior chemotherapy or immunotherapy. A 2-stage Simon-type phase 2 study design31 was used to minimize the expected sample size if the response rate was low. A response rate of ≥20% was targeted versus the null hypothesis of a 5% response rate. The original plan was for 12 patients to be accrued in the first stage. If there were no responses, the study was to be terminated. If ≥1 responses were noted, an additional 25 patients were to be accrued for a total sample size of 37 patients. The primary data analysis was estimation with 95% confidence intervals (95% CIs) of the objective response rate. The response rates were calculated for patients with measurable disease. For the analysis of secondary objectives, toxicities were summarized by grade, frequency, and duration. Time to disease progression and survival time were estimated by the Kaplan-Meier method.32


Patient Characteristics

A total of 21 patients with AJCC stage IV melanoma were enrolled between October 2003 and May 2005, and all patients met protocol eligibility requirements. Among these patients, 5 were female and 16 male and the age range was 33 to 82 years (median, 64 years). Twenty patients had primary cutaneous melanoma and 1 had primary choroidal melanoma. Two patients had AJCC disease stage Mla, 6 had Mlb, and 13 had Mlc disease.33 Seventeen patients had been previously treated with chemotherapy, 11 with immunotherapy, and 3 with radiotherapy. Eleven patients had an ECOG performance status of 0 (normal) and 10 had an ECOG performance status of 1 (ambulatory). Baseline patient and disease characteristics are shown in Table 1.

Table 1. Patient Demographics and Baseline Disease Characteristics
VariableNo. of patients (%)
  • ECOG indicates Eastern Cooperative Oncology Group; AJCC, American Joint Committee on Cancer.

  • *

    A patient may have received ≥1 prior systemic therapies.

Age, y 
 Female5 (24)
 Male16 (76)
ECOG performance status 
 011 (52)
 110 (48)
Prior therapies* 
 Immunotherapy11 (52)
 Chemotherapy17 (81)
 Radiotherapy3 (14)
 None1 (5)
AJCC stage 
 M1a2 (10)
 M1b6 (28)
 M1c13 (62)
Melanoma type 
 Cutaneous20 (95)
 Choroidal1 (5)

Treatment Details

Four patients did not start or complete the first cycle (8 weeks) of therapy. Among 17 patients who completed at least 1 cycle, 6 received 1 cycle, 7 received 2 cycles, 1 received 3 cycles, 1 received 4 cycles, and 2 received 5 cycles. There were no dose modifications. The most common reason for treatment discontinuation was disease progression (16 patients; 76%). For 1 patient, treatment was discontinued because of the development of idiopathic thrombocytopenic purpura; 1 patient because of mental status changes and fatigue; 2 patients after worsening performance status, 1 of whom suffered a fall; and 1 patient had a fall and hip fracture requiring surgical repair as well as a hernia repair that delayed his treatment and was therefore taken off protocol.


Among 17 patients completing at least 1 cycle and considered evaluable for response, 1 (6%) had a PR lasting 7 months (14th evaluable patient) and 10 patients (59%) had SD for at least 6 weeks after 1 cycle, with a range of 1.5 to 12 months, but all eventually developed progressive disease. The objective response rate was 6% (95% CI, 0–29%). Six patients had disease progression after completing 1 cycle of therapy. The patient with a PR had M1c disease. The 10 patients with SD had stage M1a (2 patients), M1b (2 patients), and M1c (6 patients) disease, and 1 of these patients had choroidal melanoma with SD lasting 6 weeks (Table 2).

Table 2. Efficacy
 No. of patients (%) (n = 17)Response duration, months
  • NA indicates not available.

  • *

    American Joint Committee on Cancer (AJCC) stage M1c disease.

  • AJCC stage M1a disease (2 patients), M1b disease (2 patients), and M1c disease (6 patients). One patient had choroidal melanoma.

Overall response rate1 (6)NA
Complete response0NA
Partial response1 (6)*7
Stable disease10 (59)1.5–12
Progressive disease6 (35)NA

Among 17 patients enrolled in the study and evaluable for response, a total of 15 patients had died at the time of last follow‒up and 2 were alive at 15 months and 38 months, respectively. The median time to disease progression was 17 weeks (95% CI, 11–38 weeks) and the median survival was 13 months (95% CI, 12–26 months). Figures 2 and 3 show the Kaplan-Meier plots of the probability of progression-free survival (PFS) and overall survival (OS), respectively. The probability of 12-month OS was 0.59 (95% CI, 0.40–0.88). The probability of 6-month PFS was 0.29 (95% CI, 0.14–0.61).

Figure 2.

Kaplan-Meier plot of the probability of progression-free survival. Dotted lines are the 95% confidence interval (95% CI) bands. All 17 evaluable patients eventually developed disease progression with a median time to disease progression of 17 weeks (95% CI, 11–38 weeks).

Figure 3.

Kaplan-Meier plot of the probability of overall survival. Dotted lines are the 95% confidence interval (95% CI) bands. Vertical tick marks at 15 months and 38 months denote times for 2 patients who were alive at the time of last follow-up. The median overall survival time was 13 months (95% CI, 12–25 months).


Table 3 summarizes the most frequently reported adverse events by severity. The most frequently reported adverse events were fatigue (81%), anemia (76%), nausea (48%), leukopenia (33%), edema (29%), dyspnea (24%), thrombocytopenia (19%), fever (19%), diarrhea (19%), increased bilirubin (19%), vomiting (14%), rash (14%), and weight loss (14%). Hematologic grade 3 of 4 adverse events included 1 case of idiopathic thrombocytopenic purpura. The most frequent nonhematologic grade 3 of 4 adverse events were high lactate dehydrogenase levels (LDH) (1 patient [5%]; unlikely related to ATO), hypoxia (1 patient [5%]), and fatigue (1 patient [5%]). There were no grade 3 of 4 cardiac arrhythmias. There was 1 case of grade 1 QTc interval prolongation. Of the 19 deaths reported as of January 2007, none was related to treatment toxicity.

Table 3. Summary of Adverse Events by Severity
Hematologic typeAll gradesGrade 3/4
No. of patients%No. of patients%
  • QTc indicates QT interval corrected for heart rate; LDH, lactate dehydrogenase.

  • *

    Idiopathic thrombocytopenic purpura.

Nonhematologic Type    
 Chills, rigors14.7600
 Weight loss314.2900
 Prolonged QTc (>0.48 s)14.7600
 Increased creatinine29.5200
 Increased bilirubin419.0500
 Increased LDH523.8114.76
Infection (without neutropenia)29.5200
 Fatigue, lethargy1780.9514.76


This clinical trial evaluated the safety and efficacy of ATO in patients with advanced metastatic melanoma. The study enrolled 21 patients, 20 of whom had previously received at least 1 prior chemotherapy and/or biologic/radiotherapy for metastatic disease. One patient had choroidal melanoma and the rest had primary cutaneous melanoma. Among the 17 evaluable patients who completed at least 1 cycle of therapy, 1 patient had a PR lasting 7 months and 10 patients (59%) had SD lasting 1.5 to 12 months (1.5 months, 1.5 months, 2 months, 2 months, 3.5 months, 3.7 months, 5 months, 6 months, 7 months, and 12 months, respectively). In a similar phase 2 study of ATO conducted at the University of Texas M. D. Anderson Cancer Center (MDACC),34 no responses were observed and 40% of patients (8 of 20 patients) had SD lasting at least 6 weeks. Of note, 50% of the patients on the MDACC study of ATO had metastatic melanoma of choroidal origin, whereas only 1 patient was accrued with ocular melanoma in the current study. This may account for the worse results observed in the MDACC study. Another difference is the treatment regimen, for which we chose a slightly higher loading dose and lower maintenance dose for a total course of 7 weeks of treatment compared with a 6-week course in the MDACC study (Pittsburgh regimen: a loading dose of ATO [Trisenox] at 0.32 mg/kg/day for 4 consecutive days in Week 1, followed by a maintenance dose of 0.25 mg/kg/day twice per week for 6 weeks; in the MDACC regimen, a loading dose of ATO [Trisenox] of 0.25 mg/kg/day for 5 consecutive days in Week 1 was used, followed by a maintenance dose of 0.35 mg/kg/day twice per week for 5 weeks). Overall, one may conclude that ATO has insufficient activity as a single agent in metastatic melanoma but may be enhanced in combination with other agents. This level of modest activity as a single agent in metastatic melanoma is not unusual when compared with other novel therapeutic agents such as the BRAF inhibitor BAY 43-9006,35 the Bcl-2 antisense compound Oblimersen, or the monoclonal immunoglobulin (Ig) G1 antibody targeting the human integrin αvβ3 receptor MEDI-522. Therefore, it is reasonable to further explore the efficacy of ATO in phase 1/2 clinical trials in combination with either chemotherapeutic agents such as DTIC or temozolomide (TMZ), or in combination with novel agents that target the main signaling and antiapoptotic pathways. Such combinations may utilize bcl-2 antisense therapy or proteasome inhibition aimed at augmenting the activity of ATO along these pathways.20, 21, 26, 28 Preclinical evidence supports the combination of ATO with proteasome inhibitors36 and ascorbic acid,37 and such a combination has been tested in hematologic malignancies.38 A phase 2 study combining ATO and ascorbic acid for metastatic melanoma is ongoing.34

Combinations of ATO with other agents are feasible, given the good safety profile that has been associated with ATO as a single agent herein. Hematologic grade 3 of 4 adverse events included only 1 case of idiopathic thrombocytopenic purpura, which may or may not have been related to ATO. Only 3 other grade 3 of 4 adverse events were noted and were nonhematologic, including a high LDH level (1 patient [5%]), hypoxia (1 patient [5%]), and fatigue (1 patient [5%]). There were no grade 3 of 4 cardiac arrhythmias, and only 1 case of grade 1 QTc interval prolongation was reported.

Extensive clinical research in the field of metastatic melanoma therapy has led to the testing of many therapeutic approaches, including chemotherapy, biochemotherapy, nonspecific immune adjuvants, cancer-specific vaccines, cytokines, monoclonal antibodies, specific immunostimulants, and novel targeted therapeutic agents. Despite these continuing efforts, new therapies over the past 25 to 30 years have not improved overall survival. The demonstrated limited impact of chemotherapy on melanoma has been paralleled by increasing understanding of tumor cell drug resistance and apoptosis resistance, for which specific molecular mechanisms of progression must be targeted. The future of melanoma therapy will ideally build on this knowledge, in which the optimal therapeutic strategy may be a combination targeting major signaling and antiapoptotic pathways.