SEARCH

SEARCH BY CITATION

Keywords:

  • tasisulam;
  • melanoma;
  • clinical trial;
  • phase 3 chemotherapy;
  • LY573636;
  • paclitaxel

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

Tasisulam sodium (hereafter referred to as tasisulam) is a novel, highly albumin-bound agent that demonstrated activity in a phase 2 melanoma study.

METHODS

In this open-label phase 3 study, patients with AJCC stage IV melanoma received tasisulam (targeting an albumin-corrected exposure of 1200-6400 h (hour).μg/mL on day 1) or paclitaxel (80 mg/m2 on days 1, 8, and 15) every 28 days as second-line treatment.

RESULTS

The study was placed on clinical hold after randomization of 336 patients when a safety review indicated an imbalance of possibly drug-related deaths in the tasisulam arm. Efficacy results for tasisulam versus paclitaxel revealed a response rate of 3.0% versus 4.8%, a median progression-free survival of 1.94 months versus 2.14 months (P = .048), and a median overall survival of 6.77 months versus 9.36 months (P = .121). The most common drug-related grade ≥3 laboratory toxicities (graded according to Common Terminology for Adverse Events [version 3.0]) were thrombocytopenia (18.9%) for patients treated with tasisulam and neutropenia/leukopenia (8.7%) among those receiving paclitaxel. There were 13 possibly related deaths reported to occur on the study, with the majority occurring during cycle 2 in the setting of grade 4 myelosuppression, all in the tasisulam arm. Investigation of the unexpectedly high rate of hematologic toxicity revealed a subset of patients with low tasisulam clearance, leading to drug accumulation and high albumin-corrected exposure in cycle 2.

CONCLUSIONS

Although the study was stopped early because of safety issues in the tasisulam arm, tasisulam was considered unlikely to be superior to paclitaxel, and paclitaxel activity in the second-line treatment of melanoma was much lower than expected. The toxicity imbalance was attributed to an unexpectedly low tasisulam clearance in a subset of patients, underscoring the importance of pharmacokinetic monitoring of compounds with complex dosing, even in late-phase studies. Cancer 2014;120:2016–2024. © 2014 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Although comprising only approximately 4% of skin cancer cases, melanoma is responsible for approximately 80% of all skin cancer deaths.[1, 2] Targeted agents, such as BRAF inhibitors, can be used in approximately one-half of patients, most of whom will develop resistance after a few months.[3, 4] Only 20% of patients might benefit from immunotherapy.[5] Thus, effective therapies for patients with metastatic melanoma remain a critical unmet need.

In a phase 2 study of tasisulam, a novel acylsulfonamide compound with a unique dual mechanism of action involving mitotic catastrophe and antiangiogenesis,[6, 7] 8 of 68 patients with second-line metastatic melanoma achieved a partial response, for an overall response rate (RR) of 11.8%.[8] The median progression-free survival (PFS) was 2.6 months and the median overall survival (OS) was 9.6 months. Tasisulam was generally well tolerated, with thrombocytopenia reported to be the predominant treatment-related grade 3/4 toxicity (20.6% of patients).

The crucial need for new therapeutic agents and the encouraging results with tasisulam in phase 2 studies led to the phase 3 study described herein in which tasisulam was compared with paclitaxel as second-line treatment in patients with metastatic melanoma. However, complications of myelosuppression occurred at a higher frequency than expected in the tasisulam arm, resulting in early study closure. Nonetheless, enrollment was sufficient to provide insight into the activity and safety profiles of each agent in the second-line treatment of metastatic melanoma, which is particularly important for paclitaxel given that it is widely used for this indication, although to the best of our knowledge, relatively little information regarding its safety and efficacy is known. In addition, pharmacokinetic (PK) analyses were conducted to explore the cause of the unexpectedly high rate of myelosuppression-related complications observed in the tasisulam arm.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Patient Population

Patients were required to be aged ≥ 18 years with a diagnosis of AJCC stage IV metastatic melanoma. Patients with active central nervous system metastasis, primary ocular melanoma, or mucosal melanoma were excluded. All patients were required to have evaluable disease as defined by Response Evaluation Criteria in Solid Tumors (version 1.0),[9] an Eastern Cooperative Oncology Group performance status of 0 to 1, and disease progression after 1 previous systemic treatment containing dacarbazine or temozolomide. Previous cancer therapies were discontinued at least 30 days before study enrollment. Immunotherapy or antibody-based regimens or treatment with a targeted agent (eg, BRAF, c-KIT inhibitor) were not considered previous treatments, unless either agent was combined with a cytotoxic drug. Baseline laboratory analytes were assessed to ensure that patients had adequate bone marrow reserve, hepatic function, and renal function.

Tasisulam is highly albumin-bound (> 99.7%); therefore, a serum albumin level ≥ 3.0 g/dL was required. Because of potential drug interactions, patients receiving warfarin (or structurally similar analogues) or those unable to withhold dosing of nonsteroidal antiinflammatory drugs or proton pump inhibitors (PPIs) for ≥ 72 h before and after treatment with tasisulam were excluded. The study was approved by local ethical review boards and was conducted according to applicable laws and the Declaration of Helsinki. All patients provided written informed consent.

Treatment

Patients were randomly assigned 1:1 to treatment with tasisulam or paclitaxel. Stratification factors included the following: lactate dehydrogenase (LDH) (baseline LDH; ≤ 240 U/L vs > 240 U/L [the upper limit of normal]), previous immunotherapy (≥ 1 course without other agent vs no prior immunotherapy, or ≥ 1 course with other agent), disease stage (M1a or M1b vs M1c), age (< 60 years vs ≥ 60 years), and sex. At the time of study initiation, there were only 2 treatments approved by the US Food and Drug Administration for melanoma, dacarbazine and interleukin-2,[10, 11] and no approved second-line therapy. Paclitaxel was chosen as the comparator because of its single-agent activity and use as a comparator in phase 2/3 metastatic melanoma studies.[12, 13]

Tasisulam (Eli Lilly and Co., Indianapolis, Ind) was administered as a 2-h intravenous infusion on day 1 of a 28-day cycle. The first tasisulam dose-escalation study used a flat-dose escalation schema up to 2400 mg, and the maximum tolerated dose was 2000 mg.[14] However, the large interpatient and intrapatient variability in peak plasma concentration (Cmax) and the high albumin binding suggested that flat dosing would not provide consistent drug exposure. Furthermore, controlling Cmax reduced the risk of toxicity, and Cmax was negatively correlated with lean body weight (LBW). Thus, the dosing regimen was revised using an LBW-based algorithm that targeted specific Cmax concentrations. Cmax-based dosing was stopped at 420 μg/mL because of dose-limiting toxicity (primarily hematologic toxicity). This was the dose used as the loading dose for the phase 2 study in patients with metastatic melanoma.[8] Meanwhile, PK, safety, and efficacy results of other phase 2 studies of tasisulam identified an albumin-corrected exposure (AUCalb) range (1200-6400 h.μg/mL) that was associated with efficacy and a more moderate risk of hematologic toxicity (data on file; Eli Lilly and Co.).[15] Thus, in the current phase 3 study, the specific dose administered to an individual patient was calculated to target this AUCalb range using an LBW algorithm and predose serum albumin level. Patients received a loading dose at cycle 1 and a lower chronic dose (65%-90% of the loading dose) because of the long tasisulam half-life (approximately 11 days-12 days). Paclitaxel (Bristol-Myers Squibb Company, Princeton, NJ) was administered at a dose of 80 mg/m2 on days 1, 8, and 15 of a 28-day cycle.

Dose adjustments were allowed after cycle 1 for patients who developed unacceptable toxicity. The tasisulam dose was reduced to 75% of the calculated dose if a patient experienced grade 4 reduced neutrophils, ≥ grade 3 reduced platelets, or any grade 3 nonhematologic adverse event (AE) (excluding allergic reactions, nausea, vomiting controlled by medication, or electrolyte imbalance controlled by supplementation) in the previous cycle. The paclitaxel dose could be reduced to 70 mg/m2 for patients who experienced grade 3/4 toxicity and, if necessary, to 60 mg/m2 after a second episode of grade 3/4 toxicity.

Assessments

The primary efficacy measure was OS. For patients not known to have died as of the data cutoff date, OS was censored at the time of last contact. A secondary efficacy measure was PFS. For patients not known to have died or to have experienced objective disease progression, PFS was censored at the time of the last complete lesion assessment. PFS was assessed at every other cycle, at the time of treatment discontinuation, and approximately every 60 days thereafter until disease progression. Best overall response was determined by the investigator according to Response Evaluation Criteria in Solid Tumors.[9]

A physical examination and recording of ECOG performance status were performed at baseline, at every cycle, and at the time of treatment discontinuation. Radiological imaging was performed at baseline, at the end of every other cycle, and at the time of treatment discontinuation. Treatment-emergent AEs and laboratory tests were collected at baseline, at every cycle, and at the time of treatment discontinuation. AEs were captured as actual terms coded to Medical Dictionary for Regulatory Activities terms. AE severity was assessed by the investigator before each visit (until at least 30 days after treatment discontinuation) according to the Common Terminology for Adverse Events (version 3.0).[16] Treatment-related toxicity was based on investigator assessment.

Statistical Analysis

The original accrual goal was 400 patients per treatment arm. The study was designed to have 80% power to achieve statistical significance at a 1-sided α level of .025 if the true hazards ratio (HR) of tasisulam versus paclitaxel was 0.75. When the study was stopped, a futility analysis based on PFS (log-rank P value > .208 to stop the trial) was performed to determine whether additional patient follow-up was warranted.

Efficacy analyses included data from all randomized patients and safety analyses included data from all patients who received at least 1 dose of study drug. Efficacy and safety analyses used intent-to-treat methodology. Kaplan-Meier analyses were performed for PFS and OS.[17] Differences between treatment arms were tested using a stratified log-rank test (see stratification factors above). Tumor RR was compared based on an unadjusted, normal distribution approximation for the difference in rates. Between-treatment differences in safety-related outcomes were compared using an unadjusted Fisher exact test. All tests of treatment effects were conducted at either a 1-sided alpha level of .025 or a 2-sided alpha level of .05. All confidence intervals (CIs) were given at a 2-sided 95% level.

PK Analyses

Plasma samples were collected during cycle 1 on days 1 (preinfusion, end of infusion [EOI], and 2-3 h after EOI), 8, and 15 and during cycle 2 on day 1 (preinfusion and EOI). Tasisulam concentrations were analyzed using a validated liquid chromatography/mass spectrometry/mass spectrometry method by Advion BioServices, Inc (Ithaca, NY).[14]

The population PK data set was analyzed using the nonlinear mixed-effect modeling program NONMEM (ICON Development Solutions, Ellicott City, Md.) (version 6) and first-order conditional estimation with interaction. A 2-compartment model was used for the PK analysis. The model selected for the total drug concentration was derived from the results of previous analysis of phase 1 and phase 2 data.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Early Study Termination and Patient Characteristics

The study was initiated in December 2009 (Eli Lilly and Co. trial H8K-MC-JZAO; ClinicalTrials.gov Identifier: NCT01006252) and conducted by 98 investigators at 92 sites in 16 countries.

An imbalance in possibly drug-related deaths was identified by the external data monitoring committee after 336 patients had been enrolled. Preliminary review of the deaths indicated that complications of tasisulam-related myelosuppression, particularly fever and neutropenia, infection, and/or sepsis, had occurred at a higher frequency than expected for this patient population, leading to the study being placed on full clinical hold. An analysis of PFS in all patients enrolled before the clinical hold failed the futility test boundary, indicating it was unlikely that tasisulam would be superior to paclitaxel if the remaining patients were followed until disease progression or death. Thus, all patients receiving study treatment were discontinued and the study was permanently closed.

At the time of study closure, 336 of the 406 patients who entered the study had been randomized, with 168 patients randomized to each treatment arm. Of the randomized patients, 325 received treatment (164 patients with tasisulam and 161 with paclitaxel), and 11 patients (4 in the tasisulam arm and 7 in the paclitaxel arm) did not receive treatment. The median age of the patients was 60 years, and the majorities of patients were white men, had an Eastern Cooperative Oncology Group performance status of 0, disease stage of M1c, an LDH level ≤ 240 U/L, and had not received prior single-agent immunotherapy (Table 1).

Table 1. Baseline Patient Demographic and Clinical Characteristicsa
CharacteristicTasisulam (N = 168)Paclitaxel (N = 168)
  1. Abbreviations: ECOG, Eastern Cooperative Oncology Group; LDH, lactate dehydrogenase; M1a, distant skin, subcutaneous, or lymph node metastasis with normal LDH; M1b, lung metastasis with normal LDH; M1c, all other visceral or any distant metastasis with normal LDH or any M classification with elevated LDH.

  2. a

    Data are presented as number (%) unless otherwise specified.

  3. b

    Totals are 167 in the tasisulam group, 168 in the paclitaxel group, and 335 total.

  4. c

    Two patients with an ECOG performance status of 2 were treated with tasisulam (protocol violations).

Median age (range), y60.04 (22-89)60.09 (24-88)
Sex  
Male103 (61.3)105 (62.5)
Female65 (38.7)63 (37.5)
Race  
White158 (94.0)162 (96.4)
Asian10 (6.0)5 (3.0)
ECOG performance statusb  
0109 (65.3)104 (61.9)
1 or 2c58 (34.7)64 (38.1)
Disease stage  
M1a or M1b58 (34.5)60 (35.7)
M1c110 (65.5)108 (64.3)
Baseline LDH ≤240 U/L95 (56.5)97 (57.7)
No prior single-agent immunotherapy142 (84.5)135 (80.4)

Efficacy

The median follow-up for all randomized patients was 5.3 months. No statistically significant differences were observed with regard to OS (median, 6.77 months for tasisulam vs 9.36 months for paclitaxel [log-rank P = .121]; tasisulam vs paclitaxel: HR, 1.23; 95% CI, 0.89-1.69 [P = .215]) (Table 2). However, PFS was found to be statistically significantly greater in the paclitaxel arm (median, 1.94 months with tasisulam and 2.14 months with paclitaxel [log-rank P = .048]; tasisulam vs paclitaxel: HR, 1.30; 95% CI, 1.01-1.66 [P = .038]). Patients with baseline LDH ≤ 240 U/L had significantly better OS (HR, 0.42; 95% CI, 0.29-0.59 [P < .001]) and PFS (HR, 0.64; 95% CI, 0.49-0.83 [P < .001]) than patients with a baseline LDH > 240 U/L. None of the other cofactors, including ECOG performance status and M1 substage, were found to have a significant effect on OS or PFS. Kaplan-Meier curves of OS and PFS are shown in Figures 1 and 2, respectively. No significant differences were observed with regard to best overall RR (Table 2). No patient achieved a complete response (CR).

Table 2. Efficacy Measures
MeasureTasisulam (N = 168)Paclitaxel (N = 168)PHR (95% CI) Tasisulam Versus PaclitaxelP
  1. Abbreviations: 95% CI, 95% confidence interval; HR, hazards ratio; NA, not applicable; NR, time not yet reached.

  2. a

    Determined using the log-rank test.

  3. b

    Includes partial responders only; no patient achieved a complete response.

  4. c

    Determined using the Fisher exact test.

Overall survival     
No. of patients with events (%)87 (51.8)68 (40.5)NANANA
Median (95% CI), mo6.77 (5.88-8.28)9.36 (6.90 to NR).121a1.23 (0.89-1.69).215
Progression-free survival     
No. of patients with events (%)139 (82.7)128 (76.2)NANANA
Median (95% CI), mo1.94 (1.87-2.04)2.14 (1.91-2.96).048a1.30 (1.01-1.66).038
Best overall response rate, no. (%)b5 (3.0)8 (4.8).396cNANA
image

Figure 1. Kaplan-Meier plot of overall survival is shown in months for all randomized patients treated with tasisulam (solid line; censored: •) or paclitaxel (dashed line; censored: ×). Overall survival is the duration from randomization to death.

Download figure to PowerPoint

image

Figure 2. Kaplan-Meier plot of progression-free survival is shown in months for all randomized patients treated with tasisulam (solid line; censored: •) or paclitaxel (dashed line; censored: ×). Progression-free survival is the duration from randomization to the date of first observation of either objectively determined progressive disease or death from any cause.

Download figure to PowerPoint

Extent of Exposure

The median number of cycles per patient was 2 for both the tasisulam arm (range, 1 cycle-9 cycle) and paclitaxel arm (range,1 cycle-12 cycles), with 25.0% of patients in the tasisulam arm and 27.3% of patients in the paclitaxel arm completing ≥ 4 cycles. The percentage of patients with ≥ 1 dose delay was 15.9% in the tasisulam arm versus 22.4% in the paclitaxel arm, the percentage with ≥ 1 dose reduction was 7.9% in the tasisulam arm versus 9.3% in the paclitaxel arm, and the percentage with ≥ 1 paclitaxel dose omission was 19.9%. The most common reason for dose reduction was thrombocytopenia in the tasisulam arm (5.5%) and neutropenia in the paclitaxel arm (5.0%). Otherwise, the reasons for dose adjustments were infrequent and comparable between treatment arms. The median tasisulam dose was 2295 mg (range, 1700 mg-3039 mg; n = 160 patients) in cycle 1 and 1920 mg (range, 1199 mg-2426 mg; n = 128 patients) in cycle 2.

Safety

The percentage of patients who reported ≥ 1 possibly drug-related AE was comparable between treatment arms (76.8% in the tasisulam arm vs 73.9% in the paclitaxel arm; P = .607). Of possibly drug-related AEs occurring in ≥ 10% of patients, alopecia was observed significantly more frequently in the paclitaxel arm, whereas thrombocytopenia was observed significantly more frequently in the tasisulam arm (both P < .001) (Table 3).

Table 3. Possibly Drug-Related TEAEs (Clinical and Biological)
TEAE/ParameterTasisulam (N = 164)

No. (%)

Paclitaxel (N = 161)

No. (%)

  1. Abbreviations: CTCAE, Common Terminology Criteria for Adverse Events; TEAE, treatment-emergent adverse events.

  2. a

    Determined using the Fisher exact test; P<.001 for tasisulam versus paclitaxel.

Any grade (≥10% in either arm)
Fatigue30 (18.3)39 (24.2)
Alopecia14 (8.5)a49 (30.4)
Nausea19 (11.6)24 (14.9)
Diarrhea22 (13.4)17 (10.6)
Thrombocytopenia37 (22.6)a2 (1.2)
Neutropenia19 (11.6)19 (11.8)
CTCAE grade ≥3 (≥2 patients in either arm)
Cardiac, general2 (1.2)0 (0.0)
Diarrhea5 (3.0)2 (1.2)
Fatigue1 (0.6)6 (3.7)
Febrile neutropenia5 (3.0)0 (0.0)
Infection2 (1.2)0 (0.0)
Sensory neuropathy1 (0.6)2 (1.2)
Biological changes CTCAE grade ≥3
Thrombocytopenia31 (18.9)3 (1.9)
Neutropenia/leukopenia23 (14.0)14 (8.7)

More patients in the tasisulam arm experienced possibly drug-related serious adverse events (SAEs) than in the paclitaxel arm (23.2% vs 8.7%; P < .001). The most common possibly drug-related SAE in the tasisulam arm was thrombocytopenia (11.0% vs 0.6% for paclitaxel). Neutropenia/leukopenia was also a more common SAE in the tasisulam arm (6.7% vs 1.2% in the paclitaxel arm).

The tasisulam arm had a significantly higher percentage of patients who discontinued treatment due to a possibly drug-related AE (8.5% vs 1.9% in the paclitaxel arm; P = .011) or possibly drug-related SAE (8.5% vs 0.6% in the paclitaxel arm; P < .001).

All the possibly drug-related nonlaboratory toxicities of ≥ grade 3 in the paclitaxel arm were of grade 3 and occurred in cycle 2 or later (Table 3). In the tasisulam arm, diarrhea and febrile neutropenia occurred predominantly in cycles 1 and 2, whereas all other toxicities occurred only in cycle 3.

Laboratory Parameters

More patients in the tasisulam arm (51.2%) experienced a grade ≥ 3 laboratory hematologic toxicity than in the paclitaxel arm (31.1%), mainly thrombocytopenia (18.9% vs 1.9%) and combined neutropenia/leukopenia events (14.0% vs 8.7%) (Table 3).

Deaths

Of the 37 total deaths reported on study or within 30 days of treatment discontinuation (24 in the tasisulam arm and 13 in the paclitaxel arm), 13 occurred in the setting of possibly related AEs (all tasisulam). The majority of these deaths were associated with complications of myelosuppression (ie, sepsis or fever and neutropenia), and most occurred in cycle 2.

PK Analyses

PK analyses explored the cause of the unexpectedly high rate of myelosuppression and related complications in the tasisulam arm. The percentage of patients with an AUCalb within the intended target range (1200-6400 h.μg/mL) was 58% in cycle 1 and 46% in cycle 2 (Table 4), which is consistent with predictions from previous phase 2 data (data on file; Eli Lilly and Co.). Approximately 16% of patients in cycle 1 had an AUCalb of ≥ 6400 h.μg/mL, a finding that is also consistent with predicted values (20%). However, 34% of patients in cycle 2 had an AUCalb ≥ 6400 h.μg/mL (23% with ≥ 10,000 h.μg/mL), which is much higher than predicted. The rate of grade 4/5 hematologic toxicity, the dose-limiting toxicity of tasisulam,[8, 18] paralleled the AUCalb, with a relatively high rate of grade 4/5 hematologic toxicity noted among patients with an AUCalb of ≥ 6400 h.μg/mL, particularly in cycle 2. Eighteen patients experienced grade 4/5 hematologic toxicity in cycle 2, the majority with higher AUCalb values in cycle 2 compared with cycle 1, suggesting drug accumulation due to lower tasisulam clearance (Fig. 3).

Table 4. Grade 4/5 Hematologic Toxicities in Cycles 1 and 2 Within Various Tasisulam AUCalb Rangesa
AUCalb Range,b h.μg/mLCycle 1 (N = 161)Cycle 2 (N = 124)
Patients in AUCalb Range No. (%)Grade 4/5 Hematologic Toxicity EventsPatients in AUCalb range No. (%)Grade 4/5 Hematologic Toxicity Events
  1. Abbreviations: AUCalb, area under the concentration time curve above the albumin-corrected threshold.

  2. a

    Toxicities were graded according to Common Terminology for Adverse Events (CTCAE; version 3.0).

  3. b

    Target AUCalb was ≥1200-6400; toxicity rates were inferred from patients included in pharmacokinetic analysis who had evaluable pharmacokinetic and available predose serum albumin data only.

<120041 (26)125 (20)0
≥1200-640094 (58)357 (46)1
≥6400-10,00015 (9)114 (11)5
≥10,00011 (7)628 (23)18
image

Figure 3. Area under the concentration time curve above the albumin-corrected threshold (AUCalb) is shown for 18 patients with grade (Gr) 4/5 toxicity in cycle 1 or 2. Predicted cycle 2 AUCalb (Right) using a pharmacokinetic-based dose adjustment (cycle 1, day 15) was simulated using each patient's actual pharmacokinetic parameters. N indicates the number of patients.

Download figure to PowerPoint

A tasisulam concentration of 130 μg/mL on day 15 of cycle 1 (C1D15) separated the patients with and without grade 4/5 hematologic toxicity in cycle 2 (Fig. 4 Top). Patients with a concentration ≥ 130 μg/mL on C1D15 had a total tasisulam clearance of 0.0102 to 0.0166 L/h (below the population mean of 0.0251 L/h, which is similar to observations in phase 1 and 2 studies[8, 14]) and comprised a very high-risk group for grade 4/5 hematologic toxicities in cycle 2 (Fig. 4 Bottom).

image

Figure 4. Tasisulam clearance and grade (Gr) 4/5 hematologic adverse events are shown. AUCalb indicates area under the concentration time curve above the albumin-corrected threshold; C1D15, cycle 1, day 15; N, number of patients. Boxes represent the 25th and 75th percentiles, and the line within the boxes represents the median. Whiskers represent the 10th and 90th percentiles.

Download figure to PowerPoint

Analysis of Additional Variables Affecting Safety Outcomes

Using univariate analyses, the use of CYP2C19-inhibiting PPIs before the safety event significantly predicted a grade 4/5 hematologic AE (odds ratio [OR], 3.16; 95% CI, 1.30-7.67 [P = .011]). Use of CYP2C19-inhibiting PPIs at any time did not predict a fatal event (OR, 1.59; 95% CI, 0.46-5.49 [P = .464]). Age > 65 years significantly predicted a grade 4/5 hematologic AE (OR, 4.47; 95% CI, 1.84-10.85 [P < .001]) or a fatal event (OR, 7.15; 95% CI, 1.88-27.18 [P = .004]). Using stepwise logistic regression analysis, the only significant predictor for most outcomes after controlling for variables (PPI use or poor metabolizer genotype, age, and creatinine clearance) was C1D15 tasisulam concentration.

These findings are consistent with analysis of the free (unbound) tasisulam concentration, which demonstrated a lower unbound clearance than the population mean (40.11 L/h) in 27 of 28 of the patients who experienced grade 4/5 hematologic toxicities in cycles 1 or 2.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

The current study compared tasisulam, a novel investigational anticancer agent, with paclitaxel as second-line treatment for patients with metastatic melanoma. Unfortunately, tasisulam-related myelosuppression and related complications occurred at a higher frequency than expected for this patient population. Although the study was closed before the primary endpoint could be evaluated, interim analysis indicated that tasisulam was unlikely to be superior to paclitaxel in this study if additional patients were followed until disease progression or death. Several studies have reported paclitaxel activity in patients with metastatic melanoma. The RR was 15.6% when paclitaxel was administered as a short infusion[19] and 12% to 16% when it was administered as a 24-h infusion, with several patients reported to achieve a CR.[20-22] In the current study, which to the best of our knowledge is the largest published to date using single-agent paclitaxel for the treatment of patients with metastatic melanoma, the RR in second-line treatment was only 4.8% with a PFS of 2.1 months, and no patient achieved a CR. Previous studies have suggested that paclitaxel plus carboplatin may have little added benefit in the second-line treatment of patients with metastatic melanoma, but may be associated with greater hematologic toxicity.[23, 24] In the current study, paclitaxel was found to be relatively well tolerated, with alopecia (30.4%), fatigue (24.2%), and nausea (14.9%) the most common possibly related AEs. Neutropenia/leukopenia was the most common grade ≥ 3 possibly drug-related AE (8.7%), a finding that is consistent with previous findings.[25, 26]

The incorporation of PK sampling into this phase 3 study provided sufficient information to determine why severe tasisulam-related hematologic toxicity occurred at a much higher frequency than in a previous phase 2 study.[8] The majority of observed grade 4/5 cases of hematologic toxicity were explained by low tasisulam clearance, which led to drug accumulation in cycle 2 due to a very long compound half-life. Because of high-affinity albumin binding and rapid, extensive metabolism of free tasisulam in vivo, preclinical studies had failed to identify a clearance mechanism for unbound (free) tasisulam, the biologically active form. Unbound tasisulam clearance was found to be lower than the population mean (40.11 L/h) in 27 of 28 patients who experienced grade 4/5 hematologic toxicities in cycles 1 or 2 (data not shown). Exploratory analyses revealed several possible explanations for the lower tasisulam clearance, including CYP2C19 polymorphisms and the use of CYP2C19-inhibiting PPIs, which are consistent with subsequent findings that tasisulam may be metabolized by CYP2C19. However, none of these factors, including clinical factors such as age, explained the majority of observed grade 4/5 cases of hematologic toxicity in the current study, suggesting that other factors likely affect tasisulam clearance and/or albumin binding. Tasisulam concentration at C1D15 was found to be the best predictor of serious hematological toxicity in cycle 2 or later, consistent with a simulation demonstrating that 17 of the 18 patients who experienced grade 4 hematologic toxicity in cycle 2 (94%) would have had an AUCalb < 10,000 h.μg/mL with a preemptive dose adjustment based on their C1D15 PK value (Fig. 3 Right). Thus, this parameter has been added to the revised individualized dosing calculator for ongoing tasisulam studies.

In conclusion, despite early study closure due to toxicity, PFS analysis did not suggest a benefit for tasisulam over paclitaxel. Although paclitaxel was well tolerated, its efficacy as a single agent in the second-line treatment of patients with metastatic melanoma was quite modest. The unexpected toxicity observed in the current trial, despite highly tailored dosing, illustrates the usefulness of PK sampling in late-phase global oncology studies, particularly those with complex pharmacology. Insights from the PK and safety findings of the current study have been incorporated into subsequent and currently ongoing tasisulam studies.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Dr. Hamid has acted as a paid consultant for and received honoraria for advisory board membership as well as payments for lectures including services on Speakers' Bureaus and research funding from Merck, Merck Serano, Amgen, Eli Lilly, Bristol-Meyers Squibb, MedImmune, Genentech, and GlaxoSmithKline. Drs. Ilaria, Hahka-Kemppinen, Kaiser, Turner, and Conti are employed by and stockholders in Eli Lilly and Company. Dr. Garbe has consulted for and/or received institutional grants, lecture fees, or travel support from Bristol-Myers Squibb, GlaxoSmithKline, MSD, Roche, Amgen, and Philogen. Dr. Wolter has received grant funding from Eli Lilly and Company for work performed as part of the current study. Dr. Maio has been an advisor to and/or received institutional grants or lecture fees/payment for the development of educational presentations from GlaxoSmithKline, Roche, and Bristol-Myers Squibb. Dr. Arance has acted as a paid consultant for and/or received institutional grants or lecture fees/payment for the development of educational presentations from Bristol-Myers Squibb, Roche, and GlaxoSmithKline. Dr. Hauschild has acted a paid consultant for and received honoraria for advisory board membership as well as payments for lectures including services on Speakers' Bureaus and research funding from Amgen, Bristol-Myers Squibb, Celgene, Eisai, GlaxoSmithKline, MedImmune, MELA Sciences, Merck Serono, MSD/Merck, Novartis, OncoSec, and Roche Pharma. Dr. Mohr has been an advisor to and/or received lecture fees/travel support from GlaxoSmithKline, MSD, Roche, and Bristol-Myers Squibb. Dr. Grob has acted as an advisor for GlaxoSmithKline, Roche, Merck, Bristol-Meyers Squibb, Celgene, Amgen, Meda, and Almirall.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES