Information Provided by Dr. Robert Kleiman, Chief Medical Officer & Vice President, Global Cardiology
During the development of oncologic agents, there are many cardiovascular toxicities to consider. Oncologic agents may produce direct cardiac contractile depression, myocardial ischemia, alterations in blood pressure, myocarditis, cardiac tamponade, hemorrhagic myocarditis, endomyocardial fibrosis, and brad arrhythmias. However, this article will focus on Torsade de Pointes, an uncommon and distinctive form of polymorphic ventricular tachycardia (VT) associated with QT prolongation. As shown in the telemetry recording below, torsades can be generated into ventricular fibrillation and be fatal.
We use the ECG for many things in clinical research. We look for new rhythms such as atrial fibrillation, serious bradycardias and tachyarrhythmias, new conduction abnormalities such as heart block or fascicular blocks, changes in the ST segment and T wave and new myocardial infarction patterns, but probably the hottest topic over the last five or ten years has been looking for the prolongation of the QTc interval.
The problem during drug development, and the reason for all of the commotion about QT, is that torsades is very rare – even with drugs which are the worst offenders. For Terfenadine (which started much of this concern) the incidence of torsades is only estimated to be about 1/50,000. For antiarrhythmic agents like Quinidine, the incidence of torsades is higher (may be as high as 1-2%). However, for the drugs that we are mainly concerned about, the risks of torsades are so low that traditional clinical trials simply aren’t powered to detect them and therefore, we need a surrogate marker in order to properly identify the drugs which may produce these harmful effects. We use the QT prolongation as that surrogate since all of the drugs which produce torsades also prolong the QT.
The QT interval can be difficult to accurately assess as there are many factors that can affect the measurement, such as the heart rate. In order to accurately compare QT values before and after dosing, you have to correct for the heart rate. This is because the QT decreases as the heart rate goes up and increases as the heart rate slows. The most common correction that has historically been used is the Bazett correction (QTcB = QT/ √RR). Although the Bazett correction becomes increasingly inaccurate as the heart rate increases, it is unfortunately the QT correction which is most commonly available to sites. The Fridericia Correction (QTcF = QT/3 √RR) is a better correction, particularly at higher heart rates. The best QT correction is the QTcI, an “individual correction” that uses a unique algorithm for each specific patient. However, to allow the calculation, this method takes approximately 40-50 ECGs at baseline and is thus limited in use.
Another difficulty is the normal variability of QTc over time. In normal volunteers the QTc can vary up to 75 milliseconds over the course of a 24 hour period, and even to 90ms in subjects with underlying structural heart disease throughout the course of a single day!
As a biomarker, QT is not ideal. This measurement, which can be difficult to assess, needs to be sensitive enough to detect quite small changes. The ICH-E14 regulatory guidance sets out the rules for how to evaluate QT for drug development trials. All new drugs with systemic bioavailability, regardless of therapeutic area and the preclinical profile, should have a thorough QTc/QT trial performed. Approved products brought back for a new dose, new indication, new population or route of administration should also have a thorough QTc/QT trial performed. Generally, the trial should be performed in healthy volunteers with placebo and positive control arms (to ensure assay sensitivity) and with both a therapeutic dose arm and a supratherapeutic dose arm, using multiples of the maximum expected dose, to simulate the worst case scenario. The trial needs to define a 5ms effect with a one-sided 95% confidence interval that excludes a 10ms effect using time matched controls.
However, drugs which cannot be given to normal volunteers are the exception to these rules and it is not required that such drugs undergo a standard thorough QTc trial. (The other exceptions are biologics, which generally do not directly affect the QT interval, and orphan drugs).
The current Oncology Division guidance is that it is necessary to characterize the ECG effects of a new oncologic agent to inform prescribers about potential cardiac toxicity and to suggest whether any additional monitoring would be recommended. The degree of effect could weigh on risk/benefit decisions for approval, but usually, QTc effects of oncologic agents used for otherwise lethal illnesses are a labeling issue and do not hinder drug approval.
Listed below are oncologic agents which are known to prolong the QT interval.
|ZD6474 (Vandetanib)||Thyroid Ca||AssymptomaticQTc↑|
|XL647||Multiple||Grade 3(DLT) QTc↑|
|SR271425||Multiple||Grade 1 -2 QTc↑|
|Lapatinib||HER-2+ Breast Ca||QTc↑|
Arsenic Trioxide, used in acute promyelocytic leukemia and sometimes in acute myelogenous leukemia, is the ‘granddaddy’ of all QT prolonging drugs. Its efficacy was established before any ECG data was collected and early reports on arsenic did not identify any QT issues. However, a formal Phase I trial showed a dramatic increase in QTc and a retrospective analysis was conducted to determine the degree of QT prolongation in patients treated with arsenic trioxide. The mean change from baseline was 47ms, with over a third of the subjects having a greater than 60ms increase. To put that into perspective, remember the ICH-E14 guidance is looking for basically a 5-10ms effect, so arsenic trioxide, despite having a huge QT effect, is still approved for human use.
When assessing cardiac safety in oncology patients, it’s important to consider whether a compound can be given to normal volunteers or whether there might ever be additional indications for use in less critically ill patients. When considering a non-cytotoxic agent, particularly if it has already been administered to healthy volunteers in earlier trials, one can assume that it will be necessary to perform a standard TQT Trial. On the other hand, when dealing with a cytotoxic agent, in order to meet the requirements for assessing cardiac safety, one may instead do PK-PD modeling in dose escalation studies or perform an intense substudy in Phase III. Of course, there are many confounding variables. QT results from early Phase I trials in patients may be difficult to interpret due to co-morbidities, electrolyte shifts, all of the different concomitant medicines (including some of the antiemetics commonly used), inconsistencies in various Phase I protocols or a lack of validated results and designs. All of these issues can burden planning and regulatory submissions. In addition, performing dedicated cardiac safety studies in oncology patients can have an adverse impact on patient care and clinical centers, as many of these centers are not accustomed to doing intense cardiac safety assessments.
In assessing the QT interval, we generally recommend the use of 12-lead digital holters which collect 12-lead ECGs continuously for 24-48 hours. This allows the ECGs to be extracted at particular time points which one has chosen based on the preliminary PK data. However, if necessary, one can go back and extract additional data to observe if there was a QT effect at a time point which hadn’t been anticipated up front. Additionally, we recommend the use of a central lab for standardized, manual readings. We also usually recommend triplicate ECGs at each time point and as many as 6-9 at baseline. This reduces intrasubject variability and the chance of finding a 60ms QTc “increase” that’s simply related to the subject’s normal daily QTc variability.
Generally, the ECGs should coincide with PK collection time points. One will assess each patient for the change in QTcF and each dose cohort for the mean change in QTcF. One can also assess QT/dose response and QT/concentration response on an ongoing basis as the dose is escalated. For the best baseline, we generally recommend 2-3 triplicates (6-9 ECGs) spaced 5-10 minutes apart. It is critical to have the best possible baseline data, as this data will be used for all subsequent comparisons for all post dosing time points. On Cycle 1 – Day 1, triplicate ECGs matched to PK sample collection time points are recommended (if inpatient, typically 6-8 post-dose ECG time points over 24 hours or if outpatient, 4-6 post-dose ECGs in 6-8 hours). During Cycle 1 (after Day 1) & subsequent cycles, depending on PK, additional ECGs (generally 1 –3 per time point) may be performed pre-dose and/or at Cmax on additional days.
The definition of a positive QTc signal will be a central tendency of greater than 10ms as well as PK-PD data showing greater than a 10ms increase in QTc at Cmax. Positive signals may also include outlier analyses showing greater than 15% of subjects having a greater than 60ms change in QTc or greater than 5% of subjects having a new QTc greater than 500ms.
What are the consequences of a positive QT study? The good news is, it doesn’t mean you have to cancel your development program! In the face of a positive QT effect, one will need to perform more intense ECG monitoring in Phases II-III in order to be able to give good recommendations in the drug label. This will allow clinicians using the drug to maintain adequate safety for their patients. On the other hand, a negative QTc study will generally mean that one only needs to perform routine cardiac follow-up during Phase III.
As previously mentioned, we recommend centralized measurement and interpretation of the ECGs because our experience is that site measurements and interpretations are very problematic and carry the risks of both false negative and false positive results. Sites generally tend to depend on unreliable ECG computer generated measurements. Also, sites may use different ECG machines, which may employ different algorithms for the correction of the QTc. In addition, many oncology sites are not all that experienced upfront with performing electrocardiographic monitoring, and readers may be inexperienced and inadequately trained at precise QTc measurement. To compound matters, many of these patients will have significant T wave and U wave abnormalities (very common in oncology patients) that make the QT measurement even more difficult to assess than in normal patients.
In summary, the assessment of cardiac toxicity and QT effects has become mandatory, but for agents which cannot be given to healthy volunteers, a robust Phase I or Phase III trial can be designed to adequately assess QT effects. Careful planning and careful implementation are key to the successful collection and analysis of cardiac safety assessments of oncologic therapies.