Cardiac Studies in Patients Treated with Depsipeptide, FK228, in a Phase II Trial for T-Cell Lymphoma

Purpose: The histone deacetylase inhibitor depsipeptide (FK228) has activity in patients with cutaneous or peripheral T-cell lymphoma. Electrocardiogram abnormalities, thought to be a class effect, were observed in preclinical animal studies and phase I testing and led to the incorporation of intensive cardiac monitoring in an ongoing efficacy trial.

Patients and Methods: This report summarizes the cardiac monitoring of 42 patients enrolled and treated on a phase II trial with depsipeptide. Cardiac evaluations included serial electrocardiograms to evaluate T-wave, ST segment, and QT interval effects and serial serum cardiac troponin I levels and left ventricular ejection fraction (LVEF) evaluations to exclude myocardial damage.

Results: Cardiac studies from 282 cycles and 736 doses of depsipeptide included 2,051 electrocardiograms and 161 LVEF evaluations. Although T-wave flattening (grade 1) or ST segment depression (grade 2) was observed in more than half of the electrocardiograms obtained posttreatment, these electrocardiogram abnormalities were not associated with elevation of cardiac troponin I or with altered left ventricular function. No significant changes in LVEF were observed, even in 16 patients treated for ≥6 months and regardless of prior anthracycline exposure. Posttreatment electrocardiograms had a mean heart rate–corrected QT interval prolongation of 14.4 milliseconds compared with baseline. Electrolyte replacement has been instituted to mitigate potential untoward effects.

Conclusion: The data obtained in this study show that the administration of depsipeptide is not associated with myocardial damage or impaired cardiac function. The potential effect of heart rate–corrected QT interval prolongation remains under study.

The histone deacetylase inhibitors (HDI) are a new class of antineoplastic agents being evaluated in clinical trials. In laboratory studies, these agents reverse the neoplastic phenotype and modulate gene expression, increasing p21 and cyclin E and decreasing cyclin D1 and c-myc expression (1–3). In addition to increasing histone acetylation, they also increase the acetylation of other proteins, such as p53, GATA4, and the estrogen receptor (4, 5). HDIs trigger both caspase-dependent and caspase-independent apoptosis (6–8). Furthermore, HDIs have been shown to be as much as 1,000-fold more toxic to malignant cells than to nonmalignant cells (9, 10).

Depsipeptide, (E)-(1S,4S,10S,21R)-7-[(Z)-ethylidene]-4,21-diisopropyl-2-oxa-12,13-dithia-5,8,20,23-tetraazabicyclo[8.7.6]tricos-16-ene-3,6,9,19,22-pentanone, is a fermentation product of Chromobacterium violaceum. Also referred to as FR901228, FK228, or NSC630176, depsipeptide was found to be a promising agent in preclinical models. Although limited activity was seen in patients with solid tumors enrolled in phase I trials conducted at the National Cancer Institute (NCI) and at Georgetown University, significant activity was seen in patients with T-cell lymphoma (11–13). Thus, a phase II trial of depsipeptide in patients with T-cell lymphoma was initiated, with the primary goal of determining the overall response rate.

In preclinical studies, QT interval prolongation and ST segment abnormalities were observed in Beagle dogs treated with rapid infusions of depsipeptide.⁴

At lethal doses, hemorrhages were detected in multiple organs, including the heart. Neither electrocardiogram abnormalities nor cardiac hemorrhage was observed when depsipeptide was infused over 4 hours. No cardiac lesions were detected in rats treated with lethal doses. With QT interval prolongation and ST segment abnormalities observed in preclinical and phase I studies, intensive cardiac monitoring was incorporated into the phase II trial of depsipeptide.

Cardiac disease in cancer patients is common and can be due to the malignancy itself, preexisting heart disease, or cardiotoxic chemotherapeutic agents (14). The most prominent heart-related chemotherapeutic toxicity is myocardial damage that may lead to impaired cardiac function and overt congestive heart failure. This type of cardiotoxicity has been observed in patients treated with anthracyclines or high doses of alkylating agents and has recently been referred to as type I chemotherapy-related cardiac dysfunction (15). Type II chemotherapy-related cardiac dysfunction, characterized by reversibility and lack of dependence on dose or reexposure to the agent, is not associated with myocyte damage and has been observed with the administration of trastuzumab (Herceptin; refs. 16–18). Similar effects have been reported with other monoclonal antibodies, including alemtuzumab (Campath; ref. 19). Other antineoplastic agents, such as arsenic trioxide (AsO₃, Trisenox), have been observed to affect QT interval duration (20). This effect is also observed with numerous nonneoplastic agents (21), including the serotonin receptor antagonist antiemetic agents commonly given to mitigate the side effects of chemotherapeutic agents (14).

In addition to evaluating the responsiveness of T-cell lymphoma to depsipeptide, secondary goals of the phase II trial were to study the effects of depsipeptide on cardiac function and to evaluate tolerability and toxicities in patients receiving depsipeptide for prolonged periods of time. In this report, we present data from the cardiac evaluations in this trial. These studies focused on evaluating myocardial integrity, cardiac function, changes in the corrected QT interval, and evidence of any potential dysrhythmia. Cardiac studies included serial electrocardiograms, cardiac enzymes, multiple-gated acquisition scans (MUGA), echocardiograms, baseline 24-hour Holter analysis, and telemetry monitoring during the first dose of the first cycle. A schema for the cardiac monitoring on protocol is shown in Fig. 1.

Patients with relapsed or refractory cutaneous or relapsed peripheral T-cell lymphoma were enrolled in a phase II trial evaluating the safety and efficacy of depsipeptide. The protocol (NCI 01-C-0049 or 1312) and informed consent were approved by the Institutional Review Board of the NCI. All data used in this analysis were obtained from patients who signed an informed consent and were enrolled and treated at the NIH Warren Grant Magnuson Clinical Center. Toxicities were reported using the NCI Common Toxicity Criteria, version 2.0. The Inclusion Criteria required measurable disease; an age of 18 years or older; an Eastern Cooperative Oncology Group performance status of 0, 1, or 2; and a life expectancy of >12 weeks. Required laboratory values included AGC ≥ 1,000/μL, platelets ≥ 100,000/μL, bilirubin < 1.5× the institutional upper limit of normal, aspartate aminotransferase < 3× upper limit of normal, and creatinine < 1.5× upper limit of normal. Patients with a myocardial infarction within the previous 6 months, a left ventricular ejection fraction (LVEF) below normal (<45% if done by MUGA, or <50% if done by echocardiogram or cardiac magnetic resonance imaging), a corrected QT interval of >500 milliseconds, unstable angina, or third-degree heart block (unless with pacemaker) were not eligible to enroll. When the 42 patients reviewed for this analysis were enrolled, there were no restrictions on concomitant medications. Of note, all patients were premedicated with ondansetron as depsipeptide was found to be highly emetogenic in phase I trials.

The first five patients enrolled in the protocol had treatment given on days 1 and 5 of a 21-day cycle with a starting dose of 18 mg/m², using the original schedule piloted at the NCI (13). Subsequently, the protocol was amended, and all patients were treated on days 1, 8, and 15 of a 28-day cycle, with a starting dose of 14 mg/m² (11). This change was made for improved patient tolerability. In all cases, depsipeptide was given as a 4-hour infusion. Provisions for dose reduction were made in the case of grade 3 or 4 toxicity. Alternatively, a dose increase was allowed for patients who had no serious toxicities. The protocol was later amended to mandate supplementation of electrolytes to achieve serum magnesium and potassium levels over 0.85 and 4.0 mmol/L, respectively, before administration of depsipeptide.

Electrocardiograms. Electrocardiograms were obtained before starting therapy, before the administration of each dose, within 1 hour after completion of the infusion, and on the day following treatment. An additional electrocardiogram was obtained on day 3 of the first cycle. Electrocardiograms were obtained using an HP Pagewriter XLi or a GE Marquette MAC1200 and recorded at 25 mm/s, with an amplitude of 10 mm/mV and with 60-Hz filtering. They were analyzed using Pagewriter A.04.01 electrocardiogram analysis software (Philips Medical Systems, Andover, MA). The QT interval measurement in this program is made by averaging the five longest QT intervals with a T or T′ wave amplitude of >0.15 mV. All electrocardiograms were interpreted by a single cardiologist (D.R.R.) in a blinded fashion. T-wave and ST segment abnormalities were assessed by either R.L.P. or S.E.B., and grades were based on definitions in the NCI Common Toxicity Criteria, version 2. Grade 1 toxicity was defined as nonspecific T-wave abnormalities (flattening or inversion without ST segment abnormalities), and grade 2 was defined as ST segment depression of at least 1 mm in at least two leads. If both were observed, then the electrocardiogram was assigned a grade 2 toxicity. A more precise evaluation of the electrocardiograms scored as grade 2 was carried out by D.R.R.

Heart rate–corrected QT interval analysis. The heart rate–corrected QT interval (QTc), indicating repolarization time, was calculated using Bazett's formula (QT divided by the square root of the preceding R-R interval) using the electrocardiogram machine software. QTc as calculated by Friderica's formula is the QT divided by the cubed root of the preceding R-R interval. All electrocardiograms with QTc ≥480 milliseconds were reviewed by a cardiologist (D.R.R.). A formal QTc analysis was done on electrocardiograms obtained during the first six cycles from patients treated on the day 1, 8, and 15 schedule (37 patients). Electrocardiograms from patients with an intraventricular conduction delay, defined as QRS duration >100 milliseconds, were excluded from the QTc analysis because, in that setting, the QTc interval includes a variable depolarization interval that leads to a prolonged QTc value (22). Six patients met this criterion and were omitted, leaving electrocardiograms from 31 patients included in the primary QTc analysis.

Cardiac troponin I assay. Serum creatine phosphokinase and troponin I levels were obtained before each dose and on the day after each dose. Assays were done in the Warren Grant Magnuson Clinical Center clinical laboratories using the Abbott Laboratories MEIA assay. Partially clotted samples may give false-positive results (23), and samples were frequently drawn from lines flushed with heparin. When possible, a new sample was drawn for repeat analysis.

Posttreatment echocardiography. Echocardiograms to evaluate potential wall motion abnormalities were also done on the day following the last dose of the cycle; day 6 for patients treated on days 1 and 5, or day 16 for patients treated on days 1, 8, and 15.

Evaluation of LVEF. LVEF evaluations were done at baseline, after the second cycle, and every three cycles thereafter using MUGA, echocardiogram, or cardiac magnetic resonance imaging (24, 25). The on-study and last available LVEF are included in this analysis to determine whether cumulative cardiac damage occurred. Cardiac measurements were done according to the American Society of Echocardiography guidelines (26). LVEFs from echocardiograms were evaluated by an independent reviewer in a blinded manner and were assessed using the biplane Simpson's method.

Rhythm. A 24-hour Holter monitor was obtained before initiating therapy to establish a baseline rhythm. Holter monitoring was done using the Oxford Medilog Excel 2 Management System (Oxford Instruments, Clearwater, FL), and recordings were done on a Medilog 45-series Analog/Digital Tape Recorder at 1 mm/s on a C60 cassette. A review and reanalysis was done on all recordings. Patients were monitored by telemetry following administration of the first dose until discharge on day 3.

A global statistical analysis, which compared baseline QTc with the QTc from electrocardiograms obtained on the day following treatment both within a cycle and for each of the cycles, was done using repeated measures ANOVA. Individual differences between paired values were evaluated for the statistical significance of the change using a Wilcoxon signed rank test. The overall statistical significance for the comparison of changes in worst electrocardiogram grade between two time points was determined using an exact marginal homogeneity test (27). All P values are two tailed.

Patient characteristics. Data from 42 of the first 43 patients who enrolled on this protocol and received at least one cycle of therapy at the Clinical Center of the NIH are included in these analyses; one patient found to have an intracardiac tumor after enrollment was excluded. Among 42 patients (25 men and 17 women), the median age was 56 years (range, 27-79 years). A summary of patient characteristics is presented in Table 1. These patients received a total of 282 cycles and a total of 736 doses. Sixteen patients with stable disease or partial or complete response received six or more cycles, including eight patients on protocol for >1 year. Fifteen patients received two or fewer cycles. Twenty-two patients had received prior therapy that included doxorubicin at a median dose of 300 mg/m² (range, 40-540 mg/m²). All patients had mature T-cell lymphomas; 24 patients had cutaneous T-cell lymphoma, and 18 patients had peripheral T-cell lymphomas. Observed toxicities were similar to those observed in the phase I trial and were primarily nausea and fatigue (11, 13).

Electrocardiogram evaluations. In all, 2,051 electrocardiograms from 42 patients were reviewed for this report. In relation to 736 doses of depsipeptide given, 1,877 electrocardiograms, 83% of the 2,250 planned, were obtained pretreatment, posttreatment, and the day following treatment. Among these, 649 were obtained before drug infusion; 630 electrocardiograms were obtained within 1 hour following completion of the infusion; and 598 electrocardiograms were obtained on the day following treatment. An additional 31 electrocardiograms were obtained on the third day of the first cycle. The remaining 143 electrocardiograms were obtained at unscheduled times.

T-wave (grade 1) or ST segment (grade 2) abnormalities noted on electrocardiograms related to all given doses are summarized in Table 2A. Reviewing electrocardiograms from all doses given, 22% had grade 1 and 2% had grade 2 abnormalities before infusion of depsipeptide; 48% had grade 1 and 3% had grade 2 abnormalities on electrocardiograms obtained immediately after completion of infusion. More marked electrocardiogram abnormalities were observed on the day following treatment, with 69% having grade 1 and 11% having grade 2 abnormalities. Abnormalities observed with the administration of the first dose are detailed in Table 2B. Similar results were observed when only the electrocardiograms from the 37 patients treated on the day 1, 8, and 15 schedule were evaluated (data not shown). Ascertaining the worst electrocardiogram grade observed at any time during treatment for each of the 42 patients, 45% had grade 1 and 52% had grade 2 abnormalities at some point (Table 2C). Similar results, 43% and 54%, respectively, were obtained when evaluating electrocardiograms obtained from the 37 patients treated on the day 1, 8, and 15 schedule (data not shown). Thus, the majority of electrocardiograms showed some T-wave or ST segment abnormality after administration of depsipeptide.

To determine whether the observed electrocardiogram changes had resolved at the time patients presented for their next dose or new cycle, we evaluated the electrocardiogram grade from the 37 patients treated on the day 1, 8, and 15 schedule. Upon presentation for treatment on day 8 of the first cycle, the electrocardiograms of 9 of 18 patients who had grade 1 and two of three patients who had grade 2 abnormalities after their first dose had reverted to normal (Table 3A). Among 16 patients who had no abnormalities surrounding the day 1 dose, two returned with grade 1 abnormalities (overall test for changes in grade: P = 0.0248). Similar results were observed when patients returned for day 15 treatment (Table 3A; overall test for changes in grade: P = 0.0033). We also evaluated the electrocardiogram grade at the time patients returned for the next cycle of therapy. Twenty-eight of the 34 patients returning for the second cycle of therapy had no evidence of T-wave or ST segment abnormalities; these included 18 of 21 patients who had grade 1 and 4 of 7 patients who had grade 2 electrocardiogram abnormalities in the first cycle. Six patients with persistent abnormalities had the same or improved electrocardiogram grade. One patient with a grade 1 electrocardiogram at baseline was scored as grade 2 in the first cycle and returned for cycle 2 with a grade 2 electrocardiogram. Data from later cycles are presented in Table 3B (all P < 0.0001 for each of the three comparisons) and suggest similar trends. Electrocardiograms scored as grade 2 were reviewed by a cardiologist for changes consistent with evidence of ischemia defined as 1 mm of ST segment depression 80 milliseconds after the end of the QRS segment. Forty five percent met this criteria. These observed electrocardiogram abnormalities raised the question of whether myocardial damage or dysfunction was associated with the administration of depsipeptide.

Troponin evaluations. To evaluate myocardial damage as an etiology of the electrocardiogram abnormalities, serum cardiac troponin I levels were obtained before the administration of depsipeptide (601 samples) and on the day following treatment (590 samples). As itemized in Table 4, 10 cardiac troponin I samples from eight patients were found to be elevated; three (3.4-11.3 ng/mL) were obtained pretreatment and were negative (≤0.2 ng/mL) upon repeat. Three elevated levels (2.1-3.3 ng/mL) obtained the day after therapy were repeated the same day and found to be negative (≤0.2 ng/mL). Creatine phosphokinase samples obtained in parallel with these 10 elevated cardiac troponin I samples were unchanged from pretreatment, (range, 16-200 ng/mL). None of these patients developed cardiac sequelae. We concluded that the troponin elevations in these eight patients were false-positive values, probably due to incomplete fibrin clot formation (23). Of note, troponin levels coincident with all electrocardiograms noted above as grade 2 were within normal limits.

Posttreatment global left ventricular function. To determine whether the observed electrocardiogram abnormalities were associated with acute left ventricular wall motion abnormalities that might be indicative of acute ischemia or myocardial stunning, echocardiograms were done on the day following the last dose of the cycle (day 6 for patients treated on days 1 and 5 and day 16 for patients treated on days 1, 8, and 15). None of the 145 echocardiograms obtained from 34 patients showed a change from baseline. Among 123 electrocardiograms obtained on the same day as the day 6 or 16 echocardiograms, 23 had ST segment depression; 82 showed T-wave flattening; and 18 had no abnormalities. These results further support the conclusion that the observed electrocardiogram abnormalities do not reflect a change in cardiac function.

Evaluation of left ventricular ejection fraction. To assess left ventricular dysfunction, 159 LVEF evaluations were done by MUGA, echocardiogram, or cardiac magnetic resonance imaging. All 42 patients underwent pre-protocol LVEF evaluations. Due to the different reference values and methodologies, we evaluated the data from patients who had MUGAs (19) separately from patients who had echocardiograms or cardiac magnetic resonance imagings (23), as summarized in Table 5A and B. Seven patients did not complete a full two cycles and did not undergo a follow-up cardiac evaluation. All others had at least one follow-up exam. The data in Table 5 are from the last follow-up exam to allow assessment of function after the patient had received the most depsipeptide possible. For patients with MUGA scans, the median time to last scan from the on-study scan for the 13 patients with both values was 4.7 months, with a range from 2 to 35 months. The median actual change in LVEF was +1%, with a range from −10% to +11%. The Wilcoxon signed rank P value for the change was 0.87; thus, no statistically significant difference was detected between the on-study and the last measured ejection fraction. For the patients followed by echocardiograms, the median time to last LVEF evaluation from the on-study evaluation for the 22 patients with both values was 4.2 months, with a range from 1 to 13 months. The echocardiograms were evaluated by an independent reviewer in a blinded manner. The median actual change in LVEF was −1%, with a range from −17% to +19%. The Wilcoxon signed rank P value for the change was 0.25; thus, there is no statistically significant difference between the on-study and the last measured ejection fraction.

Of note, 18 patients underwent LVEF assessment by both MUGA and echocardiogram before starting therapy. The median MUGA LVEF was 59% compared with 69% by echocardiogram, the median paired difference was −4% (−27% to +15%). Therefore, it does not seem that the two measurements are interchangeable (Spearman correlation: r = 0.33, P = 0.19). This confirms the distinct lower limits of normal used for LVEF evaluation by MUGA and echocardiogram.

Evaluation of patients treated for six or more cycles. Sixteen patients received six or more cycles of therapy (6-47), with a median cumulative dose of 390 mg/m² (range, 164-1,674 mg/m²), or 705 mg (range, 269-3,761 mg), of depsipeptide. As shown in Table 6, their LVEF evaluations did not show a change from baseline. Of these patients, 8 (50%) had prior therapy that included doxorubicin at an estimated cumulative prior doxorubicin dose of 300 mg/m² (range, 80-400 mg/m²). These eight patients also did not show a difference in pretreatment and posttreatment LVEF.

Efforts have been made to determine what absolute QTc value or change of QTc from baseline should be used to evaluate drugs in development. Recommended variables for evaluation include QTc values >450, 480, or 500 milliseconds or QTc interval increases from baseline of >30 or 60 milliseconds (28–30). We examined the QTc as calculated by Bazett's formula in the 2,051 electrocardiograms. Before starting protocol treatment, three patients had electrocardiograms with QTc >450 milliseconds (453, 454, 458). After initiation of therapy, QTc values >450 milliseconds were detected in 163 (8.0%) electrocardiograms from 28 patients, >480 milliseconds in 20 (1%) electrocardiograms from 10 patients, and >500 milliseconds in 5 (0.2%) electrocardiograms from four patients. Eighty-nine percent of electrocardiograms with QTc values >450 milliseconds were obtained either immediately posttreatment or on the day following treatment. Of note, two of the four patients with QTc of >500 milliseconds had an intraventricular conduction delay detected on electrocardiograms obtained before initiation of protocol. It has been shown that Bazett's formula overestimates the QTc at higher heart rates (31). All electrocardiograms with QTc >450 milliseconds were associated with a heart rate of ≥60 beats per minute and 81% with a heart rate of ≥80 beats per minute. When all QTc were recalculated using Friderica's formula, QTc values >450 milliseconds were detected in 27 (1.3%) electrocardiograms from 15 patients and >480 milliseconds in 3 (0.1%) electrocardiograms obtained from three patients; no QTc was >500 milliseconds.

A formal analysis was carried out to determine the statistical significance of changes of QTc from baseline using the Bazett-calculated QTc values from electrocardiograms obtained during the first six cycles of therapy in patients treated on the day 1, 8, and 15 schedule. Electrocardiograms obtained pretreatment, posttreatment, and the day following treatment were included in this analysis, as well as electrocardiograms obtained on the third day of the first cycle. Electrocardiograms from patients showing a baseline intraventricular conduction delay were excluded. The analysis comprised electrocardiograms related to 349 doses given to 31 patients. Among 1,078 expected electrocardiograms, 1,042 (96.7%) were obtained and available for evaluation including 339 obtained immediately posttreatment and 331 obtained on the day after treatment. Initial analysis revealed that there was no statistically significant difference among the QTc values obtained at the beginning of each cycle (P = 0.98, by repeated measures ANOVA). Further analysis showed that there was no statistical difference in QTc obtained before the administration of each dose within a cycle. This indicates that any change of QTc from baseline did not persist from one cycle to the next or from one dose to the next dose. The mean corrected QT interval values for all pretreatment electrocardiograms on days 1, 8, and 15 and posttreatment electrocardiograms obtained on days 2, 9, and 16 are shown in Table 7. In this analysis, the mean value for the QTc on electrocardiograms obtained on the day after treatment was 14.4 milliseconds longer than the mean value of the QTc on electrocardiograms obtained before treatment (P < 0.0001).

Because the QTc values obtained at the beginning of each cycle were not statistically different, a mean precycle QTc was determined for each patient and defined as baseline. Differences in the QTc values using this baseline for each patient are presented in Table 8, and a box plot of the absolute QTc values is presented in Fig. 2. A median increase of 16.5 milliseconds (range, −12.5 to +29.5 milliseconds; P < 0.0001, by Wilcoxon signed rank test) was observed on electrocardiograms obtained immediately posttreatment after the first dose of the first cycle and 10.8 milliseconds (range, −33.7 to +78.3 milliseconds; P < 0.0009) on the electrocardiograms obtained on day 2 of the first cycle. Similar changes in QTc were detected after treatment with the second and third doses of the cycle (data not shown) and on subsequent cycles (Table 8). No statistically significant difference was observed between the QTc on the electrocardiogram obtained immediately posttreatment and the QTc on the electrocardiogram obtained day 2 posttreatment ([D2]-[4 hours], P = 0.98). QTc values from electrocardiograms obtained on day 3 were statistically lower than those obtained on day 2 (P = 0.0009) and similar to those obtained pretreatment (P = 0.25), indicating that any changes observed with treatment and within the first 24-hour period had reverted to baseline by 48 hours after treatment. Overall, 514 of 670 electrocardiograms obtained immediately posttreatment or on the day following treatment showed an increased QTc from baseline (median, 13.42 milliseconds; range, −49.00 to +78.25 milliseconds). The maximum increase in QTc was 78.25 milliseconds, which represented an increase of 19% over that patient's mean baseline QTc of 415.75 milliseconds. The maximal increase of QTc ranged from 3% to 19%, with a median of 11.3%. In this set, 25% and 2.6 % of the doses were associated with a 30- to 60-millisecond and a >60-millisecond prolongation of QTc, respectively. Similar results were obtained when this evaluation was done on electrocardiograms obtained in the first six cycles from all 42 patients with 24% and 3.2 % of the doses associated with a 30- to 60-millisecond and a >60-millisecond prolongation of QTc, respectively.

Rhythm evaluations. Although this study was not designed to quantitate ectopy, some rhythm data are available. Pretreatment Holter monitoring was obtained in 37 of the 42 patients. Supraventricular tachycardia or ventricular tachycardia defined as more than three consecutive aberrant beats was noted in a significant percentage of patients, with supraventricular tachycardia noted in 14 (38%) patients and ventricular tachycardia noted in 5 (14%) patients with 4 (11%) patients having both. Supraventricular or ventricular ectopy was also frequently noted before initiation of depsipeptide therapy, with 24 (65%) and 14 (38%) patients found to have more than one supraventricular or ventricular ectopic beat per hour, respectively.

Holter monitoring for 24 hours was done during 20 administrations of depsipeptide to nine patients, providing additional safety information. Two patients were noted to have supraventricular tachycardia; however, these episodes were similar to those observed on the pretreatment Holter. No patient had ventricular tachycardia, including two patients observed to have ventricular tachycardia pretreatment.

Telemetry monitoring for 24 to 36 hours was done on 36 patients during and after administration of the first depsipeptide. One patient had significant ectopy before starting therapy, with ventricular tachycardia, supraventricular tachycardia, and episodes of accelerated idioventricular rhythm. This patient underwent both pretreatment and posttreatment electrophysiologic studies that showed no inducibility. This patient's ectopy was controlled with metoprolol, and depsipeptide was given without incident. Three additional patients were noted to have supraventricular tachycardia: one patient with a four-beat run, another with a five-beat run, and a third with two runs of six and seven beats. These patients were found to have frequent supraventricular ectopy, paired ectopic beats, or several runs of tachycardia, respectively, on pretreatment Holter monitors. Three additional patients were noted to have ventricular tachycardia: two patients, each with a four-beat run, and another patient with a 10-beat run. Notably, paired ventricular beats, a three-beat run of ventricular tachycardia, and two runs of up to six beats of ventricular tachycardia were detected, respectively, on pretreatment Holter monitoring in these three patients.

Cardiac events observed on trial. The one patient excluded from the present report was a 62-year-old patient who was found to have a true-positive elevated troponin after receiving the first dose of his second cycle of depsipeptide. Cardiac evaluation included a cardiac magnetic resonance imaging that detected an intracardiac mass that was later confirmed to be T-cell lymphoma. This patient was then removed from study. One of the patients found to have a QTc of >500 milliseconds was then placed on telemetry and was observed to have a 12 beat run of ventricular tachycardia that was asymptomatic and did not recur. Concurrently, this patient had abnormal magnesium and potassium levels that may have been related to lymphoma, prior chemotherapy, or underlying celiac disease. Four additional patients with ventricular tachycardia were described in the preceding section. One patient had ventricular trigeminy. This patient was noted to have significant ventricular ectopy on the pretreatment Holter monitor. Furthermore, this patient also was noted to have potassium and magnesium levels below normal, leading to a recommendation from cardiology to maintain potassium and magnesium levels in the high reference range. The decision was then made to incorporate this into the protocol for all of the patients. Another patient developed atrial fibrillation. This patient had a history of chronic obstructive pulmonary disease and premature atrial contractions on Holter monitor before treatment with depsipeptide.

This report summarizes the cardiac evaluations from 42 of the first 43 patients treated on a phase II trial of depsipeptide in patients with cutaneous and peripheral T-cell lymphoma. This agent showed activity in patients with T-cell lymphomas early in clinical development, and efforts are ongoing to confirm both activity and safety. As observed in previous studies, the administration of depsipeptide is associated with transient T-wave flattening or ST segment depression, electrocardiogram abnormalities that may represent a class effect associated with the administration of HDIs.

These abnormalities were not associated with myocardial damage, as evidenced by serum troponin and creatine phosphokinase evaluations, or with any change in cardiac function, as evidenced by echocardiograms obtained at the time of a patient's worst electrocardiogram abnormalities. Analysis of the QTc interval before and after treatment determined that a median QTc prolongation of 14.4 milliseconds followed treatment with depsipeptide. However, it is important to note that all of the patients were receiving a serotonin inhibitor for antiemetic therapy and that these agents have been associated with QTc prolongation. Moreover, any observed QTc prolongation was short-lived, and no patients had treatment-related sustained or symptomatic arrhythmia.

Antineoplastic agents, in general, have the narrowest therapeutic indices of all pharmaceutical drugs. Although they usually affect more rapidly dividing cells, such as bone marrow and the gastrointestinal tract, they may also have toxicity against nondividing cells, including neurons and myocardial cells. If sufficient, toxicity to myocardial cells can lead to the development of cardiomyopathy and congestive heart failure. In patients treated with anthracyclines, the risk of developing a cardiomyopathy is primarily associated with cumulative dose (32). Alkylating agents are also associated with a risk of cardiomyopathy, most notably when given at high doses (33). Detection of anthracycline-induced cardiomyopathy traditionally relied upon MUGA or echocardiography. Recent studies have evaluated the use of serum cardiac troponin I or T levels as a more direct method of detecting myocardial damage following chemotherapy, an approach used in the early detection of myocardial infarction (34–37). In preclinical models, the detection of serum cardiac troponin was associated with cardiac damage, with serum troponin levels correlating with dose of doxorubicin given and extent of damage observed (35, 36). In patients, detection of cardiac troponin in the serum after exposure to anthracyclines or high-dose cyclophosphamide predicted patients at greater risk of developing cardiomyopathy (34). Furthermore, patients with hematologic malignancies previously exposed to anthracyclines had higher detectable levels of cardiac troponin I than patients not previously treated with anthracyclines (37). These results give our findings greater importance because no clinically significant elevation of serum troponin was detected, even in patients who had prior anthracyclines. Of note, patients treated and followed for prolonged periods of time (up to 50 months) did not show any evidence of cardiac damage.

Aside from myocardial damage due to either direct myocyte injury, such as that seen with anthracyclines or ischemia, T-wave, or ST segment abnormalities, can also be associated with myocardial stunning. Stunned myocardium is defined as left ventricular dysfunction with normal flow (38). Echocardiograms obtained at the time of electrocardiogram abnormalities show no wall motion abnormalities, indicating that patients did not have stunned myocardium as an explanation for the observed electrocardiogram abnormalities.

Although prolongation of the QTc over 500 milliseconds is commonly accepted to increase the risk for torsades de pointes, the absolute increase in risk remains unclear (39). Some agents that have profound QTc prolongation have little risk for torsades de pointes, whereas other agents that are associated with slight QTc prolongation have a markedly increased risk (40). Furthermore, it is difficult to quantify the risk of cardiac events for any particular drug, even between individuals with equivalent degrees of QT prolongation (41). Individual risk factors proposed include gender, hypokalemia, hypomagnesemia, bradycardia, recent conversion from atrial fibrillation (especially with a QT prolonging agent), congestive heart failure, digitalis therapy, high drug concentrations, rapid rate of IV infusion with QT prolonging agent, baseline QT prolongation, subclinical long-QT syndrome, and ion channel polymorphisms (41).

Additional investigations are needed to confirm the observed changes in the QT interval. Ideally, these evaluations should follow the new recommendations of drug regulatory agencies (30). Patients should not have an elevated QTc at baseline; serum levels should be corrected for hypokalemia and hypomagnesemia; and concomitant use of agents that may prolong the QTc should be avoided. Studies may also need to control for diurnal variations as well as the use the use of concomitant medications, such as antiemetics.

Administration of arsenic trioxide IV daily was found to progressively prolong the QTc over the first 6 days of treatment, reaching a mean increase of 47 milliseconds (20). In 36.6% of courses, a 30- to 60-millisecond prolongation of QTc was documented; in 35.4% of courses, a >60-millisecond prolongation of QTc was documented. Of note, electrocardiograms with “markedly flattened T waves” were excluded from analysis, presumably due to the difficulty in measuring the QT interval in that setting. One strategy employed in the therapy with arsenic trioxide is supplementation of potassium and magnesium to high normal levels because hypokalemia and hypomagnesemia are associated with electrocardiogram abnormalities that include flattening or inversion of the T wave, U waves, ST segment depression, and prolonged QT interval. In our analysis, QTc values could be obtained from all of the electrocardiograms, at times requiring interpretation by a cardiologist. It should also be noted that abnormalities of magnesium and calcium have been previously reported in patients with advanced cutaneous T-cell lymphoma (42). Supplementation of potassium and magnesium was initiated after a patient with hypokalemia and hypomagnesemia developed a QTc interval of >500 milliseconds. We analyzed the electrolyte levels obtained before the administration of each dose of therapy and found that, for the 42 patients studied here, 39% of doses would require supplementation with either potassium or magnesium, and 17% would require both when using target levels of 4.0 mmol/L for potassium and 0.85 mmol/L for magnesium.

Nausea is one of the more common side effects of the HDIs (43). The most reliable and potent antiemetic regimens used in oncology include the serotonin receptor antagonists. Prolongation of the QTc has been noted in patients treated with serotonin receptor antagonist antiemetic agents, with some studies reporting 15 milliseconds as the mean QTc prolongation observed (reviewed in ref. 14). An effect on QTc has been noted with the use of dolasetron, ondansetron, and palonosetron but not granisetron. We are unable to dissect how much of an effect the antiemetics may have contributed to the observed QTc prolongation, or whether additivity or synergy occurred. All patients in this analysis received ondansetron. Because these agents are necessary for treatment of nausea in most patients receiving depsipeptide, we now recommend the use of granisetron in trials with depsipeptide. With the use of agents associated with QTc prolongation and also metabolized by CYP3A4, such as terfinide, it has been previously noted that QTc prolongation may be exacerbated by exposure to other drugs that inhibit CYP3A4 function (44). Some studies suggest that depsipeptide may be metabolized by CYP3A4 (45). Thus, to minimize potential untoward effects in patients, coadministration of agents known to prolong the QTc or inhibit CYP3A4 are precluded in clinical trials of depsipeptide.

Cardiac events have been reported in clinical trials with other HDIs, including electrocardiogram abnormalities, tachyarrhythmias, and QTc prolongation. Electrocardiogram effects described as nonspecific T-wave or ST segment abnormalities have been reported with the use of depsipeptide, SAHA and LAQ824 (13, 46–48). Supraventricular tachycardia, the dose-limiting event reported from the NCI phase I trial with depsipeptide, was reported as a dose-limiting toxicity in a phase I trial of PXD101 and was reported in patients treated on a phase I trial of MS-275 when given daily (13, 49, 50). In patients treated with LBH589, a hydroxamic acid, QTc prolongation of >500 milliseconds was noted in 5 of 12 (42%) patients treated at higher dose levels (51). QTc prolongation associated with LAQ824, another hydroxamic acid, was also reported (48). Of note, a statistically significant increase in QTc interval was reported with 10% of 77 patients treated having QTc prolongation of >60 milliseconds. One patient found to have QTc of >500 milliseconds experienced torsades de pointes when retreated at a lower dose level (48). The fact that effects on the QTc have been observed with HDIs with different chemical structures suggests that the observed effects are probably a class effect due to their underlying mechanism of action as opposed to a specific effect of depsipeptide. It may be possible to develop HDIs that retain cancer cytotoxicity without an effect on the QT interval.

Of note, among over 450 patients treated to date during the development of depsipeptide there have been six on-study deaths that have been reported as possibly related to depsipeptide. Five of these were acute, occurring 16 to 60 hours following depsipeptide administration. Four of the five patients had significant risk factors for sudden death (52): one patient had hypertension and chronic obstructive pulmonary disease; one patient with a neuroendocrine tumor had left ventricular hypertrophy and uncontrolled hypertension; one patient had severe atherosclerotic vascular disease requiring renal and cardiac stents; and one patient with structural heart abnormalities and left ventricular hypertrophy had developed atrial fibrillation and was taking digoxin. The fifth patient had a history of pulmonary sarcoid and was cotreated with palonosetron, a serotonin receptor antagonist antiemetic agent with a 40-hour half-life. Palonosetron is associated with QTc prolongation and metabolism by CYP3A. A sixth patient on a combination trial of depsipeptide and decitabine died 10 days after the last dose of therapy after developing symptoms suggestive of a pulmonary embolus. Whether these deaths were related to depsipeptide, to the risk factors themselves, or to a combination is not known. However, it seems prudent to exclude patients at risk for sudden death in trials using histone deacetylase inhibitors. In addition, further studies will need to evaluate the relationship between pharmacokinetics and QTc. Ongoing studies include the quantitation of ectopy by Holter monitoring before and after depsipeptide infusion.

Recently, a number of studies have linked HDACs to several pathways involved in the generation or the repression of cardiac muscle hypertrophy. Interestingly, class I and II HDACs seem to have opposing effects, one of the only differences between the two classes identified. Class II HDAC enzymes repress the activity of MEF2, a transcription factor that, when released, allows expression of a program of genes that leads to cardiac hypertrophy (53). In contrast, class I HDACs are involved in repression of an antihypertrophy program. HDIs increase the expression of this program in treated myocytes and prevent cardiac hypertrophy in mice transgenically engineered to develop it (54, 55). Notably, no evidence of significant change in cardiac muscle thickness was observed in patients treated with depsipeptide on this study as evaluated by echocardiogram (56).

Depsipeptide is a member of a new class of antineoplastic agents that seems to have great promise as part of the armamentarium in the fight against cancer. To determine whether depsipeptide was associated with serious cardiac side effects, multiple cardiac evaluations were incorporated in this trial. These studies included serial electrocardiograms, serum troponin I levels to test for evidence of myocardial damage, and evaluation of cardiac function. In this analysis, the observed electrocardiogram changes were reversible and generally of short duration. In addition, no evidence of cardiotoxicity was detected. With these studies showing no acute or chronic impairment in cardiac function, we conclude that the electrocardiogram abnormalities observed with the administration of depsipeptide are not indicative of myocardial dysfunction or myocardial damage. Additional investigations are needed to determine whether the QTc prolongation has an effect on the safety profile of this agent. These data extend and support the safety data for an agent that has shown significant clinical benefit in patients with cutaneous and peripheral T-cell lymphoma.

We thank Susan Bakke for being the first research nurse for this protocol, Drs. Susan Jewell and Lyudmila Kalnitskaya for their efforts in compiling the database, Joye Lynne Byrne for data analysis, Dr. Richard Cannon for his help as the first cardiology consultant on this protocol, and Dr. Stanislav Sidenko for his review of all of the echocardiograms obtained.