Current regulatory requirements
As per the ICH S7 guideline document, two GLP-compliant preclinical studies need to be conducted in support of an IND application, for all drug types, including biologics such as peptides and antibodies:
- A conscious-animal telemetry study monitoring ECG intervals continuously over a period of at least 24 hours post-dose. The study generally includes bioanalytical confirmation of plasma exposure levels, and dosage is determined by the anticipated clinical route of administration of the drug. Doses to be tested are required to cover and exceed the anticipated human exposure, as confirmed by the aforementioned bioanalytical work. Analysis of the ECG intervals can be done automatically by the software, with a manual over-read for any arrhythmic episodes. While QT correction methods based on Bazett’s or Fridericia are still used, the most common is the individual-regression correction of the QT interval for heart rate, in which 24-hours baseline heart and QT data is analyzed to quantify each animal’s specific QT/HR regression. That regression is then used to correct post-test article exposure data and yield a corrected QT interval, QTc.
- An in-vitro patch-clamp assay using HEK293 cells transfected with the human ERG gene, which encodes the channel allowing the IKr potassium current. The assay should be conducted by manual patch-clamp rather than automated patch-clamp and variants, and be conducted at physiological temperature to reproduce the in-vivo kinetics of activation, deactivation, and inactivation. Cumulative concentrations should be perfused over the patched cells, for sufficient time to achieve steady state current inhibition. The concentrations to be tested ideally would cover the therapeutic range of anticipated human exposure (plasma levels), and exceed them by a factor of 30-fold (USA) or 100-fold (Europe, Canada). It is understood however than anticipated plasma exposure are uncertain, and that solubility of the test article in hERG-assay compatible matrices may be limited. The guidelines (ICH S7b) specifies that concentrations to be tested should be as high as allowed by the solubility of the test article. As for in vivo studies, bioanalytical chemistry is expected to confirm levels of exposure in the hERG assay. Finally, the study design should include 1- a negative control/vehicle group to allow quantification of the effect of the solvent alone and time-dependent current run-down; 2- a positive control to demonstrate the sensitivity of the assay; 3- a reference compound, selected for its ability to inhibit IKr currents and with a clinical history of inducing mild QT interval prolongation. This reference compound does not have to be related to the test article: it is included to allow the regulators to translate the % IKr inhibition in-vitro with a QTc prolongation in the clinic.
These two assays produce complementary data (animal vs. human, in-vitro vs. in-vivo) which is integrated with PK data and included in the cardiovascular section of the IND filing. Both regulator assays above are run prior to clinical testing, generally some time before toxicity studies are initiated.
Changing regulatory landscape
The current preclinical paradigm (hERG + telemetry) is followed by a Thorough QT (TQT) study, conducted as per document ICH E14, before Phase 3 clinical trials start. The TQT study is expensive and comes at a time when significant clinical development has been completed. The upside is that the preclinical+clinical combination is extremely successful in identifying drugs with QT liability. Since the adoption of E14, there has not been a single drug withdrawal resulting from unexpected QT prolongation in the USA. On the other hand, it is believed the requirements are so stringent several drugs with therapeutic potential have had their development discontinued as a result of a positive signal in one of the three studies.
Two proposals have been produced by the pharmaceutical industry to remedy the situation, and refine our preclinical and testing schemes.
- In 2015, the FDA accepted that TQT studies be moved from before Phase-3 to Phase 1 trials, with one arm within a normal Phase 1 study cohort dedicated to ECG acquisition and QT interval measurement. The analysis is to be correlated with levels of exposure, rather than by-time, which is easy in a Phase-1 study. Already, most ongoing clinical TQT trials use the new paradigm defined as an earlier-and-refined, lower-cost and lower-risk approach. Effective now, the change in E14 will be addressed officially in 2016, in a Q&A document complementing the current E14 document. Most companies will adopt the practice rather than run a stand-alone TQT study later in development.
- In 2013, the pharmaceutical industry proposed “CIPA”, or Cardiovascular Integrated ProArrhythmic assessment, consisting of more preclinical ion channel testing using what is expected to be automatic patch-clamp and cell lines, combined with human induced pluripotent stem cells transformed into cardiac-myocyte-like cells and in silico models. The CIPA initiative – for all its sound scientific basis – has had a hard time gaining traction with the regulators: ongoing efforts to harmonize methodologies and demonstrate significant gains have thus far disappointed, and CIPA has evolved from a “regulatory framework for preclinical safety assessment” to a “promising strategy to be implemented voluntarily”.
Current trends in preclinical testing
Binding profiles for several cardiac and non-cardiac ion channels and potential receptors are regularly seen before proper safety testing is designed and implemented. Binding panels provide an indication of potential affinity, which may, or may not, translate into an actual functional impairment of the target being bound. The panel results are often used to identify which channels will be subjected to manual patch-clamp testing, and the range of concentrations to be tested.
Beyond the binding panel assays, there are two strategies and a sub-group of compounds which cover essentially 99% of the preclinical safety assessment strategies in 2015-2016.
- Smaller companies tend to rely on a conservative, minimalist decade-old approach consisting in running a hERG inhibition screen, followed by GLP-compliant hERG inhibition and CV telemetry studies (the actual regulatory safety package). In this strategy, a positive hERG inhibition signal in screening assay leads to an isolated heart assay, or a Purkinje action potential duration (APD) assay, to give context to the inhibition. The “side-step” is actually European in origin; The EMEA has often expressed the desire of seeing cardiovascular findings put in context using a so-called complementary assay, such as Langendorff isolated hearts, or extracted Purkinje fibres APD assay.
- Larger companies take a more integrated approach to cardiovascular safety, often initiating their investigation with an anesthetized guinea pig screen, in which ECG intervals are analyzed. Based on the findings, specific ion channels are selected for either automated or manual patch-clamp (manual patch-clamp seems to remain a favourite based on requests received at IPST), in an approach reminiscent of the CIPA proposal, described above. GLP-compliant hERG inhibition and telemetry QT prolongation studies (again, the regulatory package) then follow these initial screening tests.
- Any compound suspected of affecting calcium channels is generally tested in a Langendorff isolated heart assay, to simultaneously measure electrical conduction and contractility. This assay is usually conducted in parallel to multiple ion channel work, as described above.
In recent months, a few requests surfaced for multiple-electrode-array (MEA) assays. The method involves measuring an ECG-like depolarisation and repolarization signal from a sheet of induced pluripotent stem cells derived into cardiac myocytes. The strategy there is to benefit from a greater throughput than the anesthetized animal assay, while also eliminating species-specific differences. The approach has the merit of providing high-throughput for initial screening. The species-specificity argument is less significant as current CV models are so well defined that translation from one model to Man is relatively simple.
The evolution of the CV testing strategy points to less cookie-cutter approaches, and more integration of existing data to inform on the next steps in testing: mechanism of action determines which binding panel to select, which returns more information on potential affinity for targets which could impact cardiovascular function. More refined CV testing is then dictated by the growing dataset available. In this context, those sponsors which do best are those who have an electrophysiologist on board, either as full-time employee, or consultant. Significant time and resources can be saved by streamlining the testing strategy based on existing data, rather than adopting an all-encompassing, overly exhaustive testing scheme, for which there is currently no regulatory requirement, and often no scientific justification.