Robert Kent is the Phase 1 and ECG Centralization Business Development at Spaulding Clinical.
Since the adoption of ICH E14 guidance “Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs” (2005), Pharmaceutical companies engaged in development of new drugs have been required to investigate the potential for a proarrhythmic effect by means of analysis of QT/QTc data derived from ECG recordings. Usually, this has been accomplished by carrying out a Thorough QT (TQT) Study.
While the implementation of the ICH E14 guidance has been successful in allowing investigators and regulators to identify compounds that can significantly increase the risk of cardiac arrhythmias and sudden death and stop the development of these drugs, it has become apparent that the adoption of the guidance has had unintended consequences that are acting as an impediment to the successful development of new drugs. These include:
Delays in the development and consequent regulatory approval due to the guidance recommendation that the TQT study is undertaken between the completion of Phase II and the start of Phase III studies.
The QT interval itself is an imperfect biomarker for proarrhythmia. This has undoubtedly resulted in the premature discontinuation of many potentially effective drugs that in reality would not have posed a risk of severe cardiac side effects.
A consequence of these limitations, FDA, in conjunction with the Cardiac Safety Research Consortium (CSRC), has been looking at ways that the TQT study goals can be achieved more efficiently and accurately. There are two alternate proposals currently being assessed:
More robust pre-clinical testing utilizing the CiPA proposal (Comprehensive in Vitro Proarrhythmia Assay]. While this approach may show promise in theory, it is years away from being a suitable replacement for clinical testing.
Replacement of the TQT study with increased ECG collection in SAD/MAD Phase I studies in conjunction with concentration-effect modeling. A pilot study carried out with the support of the CSRC showed this approach was able to replicate the results that would be generated by a TQT study. Unfortunately, the exciting implications of this study have been somewhat marred by some controversy concerning claims about “acceptable” methodology for analyzing the ECG data. However, it is not within the remit of this posting to enter this debate except to say that there are several ways of analyzing the ECG data that may be acceptable to the relevant regulatory authorities and that one methodology is not preferred over the others.
If this approach does eventually replace the TQT study, the smaller cohort sizes typically found in SAD/MAD studies will result in a smaller number of ECG’s available to be used for analysis. Consequently, accurate prediction of QT effect will require, more than in the past, that all ECG’s collected are of the highest quality possible.
Spaulding Clinical Research looks at three key metrics to when determining the quality of ECG data collected and analyzed for the assessment of QT data; the % of readable ECGs:
Percent of Readable ECG’s
While it might seem self-evident that it is essential that readable ECG’s are collected by the investigative site at the correct time points relative to the PK/PD profile of the drug it is remarkable how infrequently this quality metric is requested. The primary factor that has a detrimental effect on the percent of readable ECG’s is the quality and surveillance of the connection between ECG device and subject/patient.
Poor quality connections result in unusable tracings. Unfortunately as a result of movement, moisture build up between skin and electrode etc. the quality of the electrode to skin contact deteriorates the longer the electrode is in place. In order to achieve the optimal level of % of readable reads, it is necessary to have well-trained site staff carry out continuous real-time surveillance of the ECG tracings. This is best undertaken at a site that has installed ECG equipment that allows both continuous collection of 12-lead ECG recording and allows real-time review for quality, such as the Mortara Surveyor system. Sites using this system have the ability to assess the ECG data quality prior to the time point and take corrective action. Other methodologies are not as reliable: standard 12-lead devices require the site staff to take the ECG’s in real-time but if there is a problem the correction has to be made and the ECG may be delayed.
The most common way that ECG’s are routinely collected for QT analysis is by utilizing 12-Lead Holter equipment. Unfortunately, a major issue with 12-Lead Holter recording is that most systems do not allow for real-time quality review; as a result the percent of readable ECGs is lower. In a related study (Salvi V et al. 2011) investigating the incidence of lead misplacement in ECG’s collected for clinical studies found that limb lead placement errors occurred in 3.4% of all the ECG’s reviewed (n=85,133) but the percent of limb lead placement errors was 7.5% in those ECG’s derived from Holter data. While limb lead misplacement does not necessarily result in an unusable ECG, this difference is illustrative of the quality problems associated with 12-Lead Holter collection and solutions based on them.
As we move to an earlier assessment of the QT interval with smaller sample sizes, data loss due to unreadable ECG’s will become less acceptable.
Heart Rate SD
Most QT correction formulas are a function of Heart Rate (HR), the more stable the HR the more stable the QTc will be. Unfortunately, the correlation between HR and QT is never strictly linear so achieving HR stability is extremely important as this maximizes the chance any variation in QTc is a drug effect and not the result of external stimulation. As HR is sensitive to multiple external influences it is imperative that the Clinical Pharmacology Unit undertaking the clinical conduct of the study makes every effort to ensure that the effects of external stimuli are minimized.
Fridericia’s QT correction (QTcF) is the standard correction formula submitted to the FDA for evaluation of potential drug-induced QT prolongation.
As the QT interval is dependent on HR, the strict control of factors affecting HR as described in point 2 are obviously important in reducing the variation in QTcF. Additionally the accuracy of the measurement is dependent on the quality of the ECG, as poor quality ECG’s often do not allow accurate determination of the end of the T wave. This applies equally to ECG data that it is interpreted by an automatic algorithm, by a cardiologist/technician or a combination of automated measurement with human review. Adequately trained personnel are essential if consistency and, thus, lower variation is to be obtained in QTcF measurements.
High-quality ECG data is dependent on the experience and processes of the investigative site collecting the data much more than on the experience of the Core ECG laboratory.
Spaulding Clinical Research, which uniquely operates a 200-bed Clinical Pharmacology Unit and an ECG Core Laboratory, has collected its data on these factors since its inception. Our metrics are shown below.