Recording and Q&A from webinar ‘In Vitro Assessment of Cardiac Risk in Drug Discovery’

Our webinar ‘In Vitro Assessment of Cardiac Risk in Drug Discovery’ enabled attendees to learn how an hiPSC-CM model can help provide clear decision-making data for their project team, avoiding costly issues related to QTc and QRS cardiac liabilities in the clinic.

Presentations were received from:

Derek Leishman (VP Translational and Quantitative Toxicology at Eli Lilly and Company) presented the opportunities a sponsor now has available to increase the efficiency and effectiveness of QTc assessment by leveraging the ICH S7B core assays. He also discussed times when the already conducted core hERG and in vivo assays do not meet the new expected ‘best practice’, and addressed how gaps in best practice or limitations in tested exposures could be mitigated.

Steve Jenkinson (VP Drug Discovery and Safety at Metrion) described a model utilizing a voltage sensitive dye in combination with a fast capture rate plate reader to generate high quality action potential recordings in a 96 well based format. Moreover, using an extensive and robust reference data set, the translation of key endpoints from this assay allows the prediction of compound plasma exposures in the clinic that will be associated with a 10 ms change in QTc for novel preclinical compounds, as well as provide an insight into the probability of potential QRS liabilities.

Questions from delegates with answers from the presenters are provided below. If you have further questions relating to this webinar, or would like to learn more about how Metrion can support you with preclinical cardiac safety please contact us.

Regarding legacy data, are sponsors currently being held to the new ICH, S7B Q&A standard? What if you have GLP hERG data that was generated using different protocols?

This is a tough one for everyone right now. It’s a situation that we’ve been in before, when IHC E14 was first introduced, we had the situation where we had a lag between the introduction of new guidance and everybody’s assays and clinical studies catching up. We’ve submitted a number of waiver requests based on legacy data. I have to say that we don’t win them all. We’ve had some, regulators have said, yeah, this hERG is fine, even though it wasn’t best practice. We’ve had others that have been more insistent on us doing a new hERG study. In general, if you have a very large margin it’s easier to argue that even if you did have some deviations from what might be best practice, given that you’ve got a large margin, it’s unlikely that the margin would erode. But you also have to have some supporting data with the reference agent. Then you’ve got the other issue that one regulator may say “Yes, I’ll accept your legacy hERG.” but another one doesn’t. So we find ourselves in the situation now of doing repeat hERGs.

If you have a compound for a life-threatening condition, for example in the oncology field where there are no current therapies, and you get a finding in E14 SB – does it still prohibit approval?

Those are situations where the question is “Do you or do you not have a risk?” And then risk benefit is the second question. We have seen that there have been a number of compounds that have been approved with QT prolongation, and if they’ve got QT prolongation, they presumably also had narrow margins in hERG and in vivo and those compounds are approved. Obviously, it depends on the indication and some of those considerations. It will always be a review question, so I don’t think you’ll get an early answer from the regulators because they want to see what the entire package looks like in the end and how good your efficacy is. It’s all very well saying it’s for a life-threatening indication, but it has to work, to really demonstrate the benefit relative to that risk. Overall, it’s a risk benefit situation – you can obviously move forward to some degree of risk.

Does the asset (hiPSC-Cardiomyocyte model) that you just presented
on actually meet the standards outlined in the updated S7B Q&As for
regulatory decision making?

When it was developed, we went through the various sections of the guidance and tried our best to address every single one of them – and I believe we addressed every single one. In the routine running of this assay it doesn’t make sense to do compound concentration verification for every study because you’re usually using this to compare a reasonable number of early compounds. This is an assay I tend to use an earlier stage in assessment not right at the very end. However, if you need to use it as part of a package, then at that stage you can do the compound concentration verification quite easily. The assay is run to the correct standards. We do have the same data that Derek alluded to (the three reference compounds Moxifloxacin, Dofetilide and Ondansetron and all have been shown to be in alignment with the FDA’s data. So, as a package, it’s very tight.

Are hERG GLP studies required for antibody therapeutics and biologics?

Generally regarding antibodies, the answer of no. There’s no requirement for a standalone study for monoclonal antibodies either. But the jury’s out on the other biologics. There’s a draft clinical pharmacology guidance from the FDA on all the nucleotides and on peptides. Both would suggest you have to do some work and treat them essentially as small molecules. You can kind of get a pass if your peptide only has natural amino acids, but my experience is that almost every modern drug peptide has some non-natural amino acid added. So that exception doesn’t really help you too much. Is an open topic though. We will go back now as the ICH S7B 14 group and develop some Q&As around this because lots of people are asking about it. ICH S7B was very much created at a time when most companies portfolios were dominated by small molecules. But actually, if you look at the portfolios of many large companies now, and some small companies focussed on biologics, small molecules have become far rarer than, than they were at the time of ICH S7B finalisation.

In terms of an E14 S7B hERG study, what about positive control?

At Metrion we obviously run these. In my experience having done electrophysiology, nobody wants to run Dofetilide due to its slow kinetics. So probably taken that one off the table. You’ve then got moxifloxacin and Ondansetron and most people’s experience is of Ondansetron being so fast is that it’s actually a pretty reasonable positive control to use. But there’s probably nothing essentially wrong with Moxifloxacin either.
Most CROs are focusing in on Ondansetron just because it’s well behaved. It’s in that sweet spot – not too potent and not too weak. So, yeah, I think Ondansetron is probably going to be the reference compound of choice. Derek is hopefully going to be releasing a publication sometime soon covering the various reference compounds and the data from multiple CROs. It’s important to have all three compounds, as at least in your back pocket, in case that’s required for a regulatory filing.

Regarding translation using IPSC: if you’ve got a compound that’s hitting multiple ion channels does that activity translate? You’ve clearly shown that it does for hERG, but what about compound with mixed ion channel pharmacology?

You’re always limited by the clinical data you’ve got. Using reference compounds is great, but it’s always interesting to see whether, it will continue to translate, for the compounds we’ve seen so far. With regards to specifically multi-ion channel effects (I can’t go into too much detail here, but this will be published sometime soon), a compound a quite potent hERG blocker, and the exposure was going to be right on top of where the hERG block was going to be. You’d expect to see QT in the clinic, but it also had a little bit of weak calcium activity, a little bit like verapamil and its profile. The QT waiver was declined on that once it had to go through to a full TQT study and the stem cell data showed that it wouldn’t have any QT prolongation and it didn’t in TQT study. We’ve got data showing that the assay will pick up defects, but we also have data for this assay showing that it will show a negative when you would expect to see a negative because of that multi-ion channel effect. So that’s reassuring.

What would it take for in vitro hiPSC-CM models to completely replace animal testing? Do you think this is possible or will animal testing not be replaceable?

Although there are several publications out there looking at various aspect of translation with respect to ephys I don’t feel that there is a sufficient number of robust studies looking at the same endpoint to really provide the confidence that we have with say manual hERG patch clamp. In addition, we do know that there are limitations to this system with some proteins not being expressed to the same level as in native tissue. That being said it could be argued that animal models suffer from similar limitations. Moreover, for functional responses (contractility) it is clear that naïve hiPSC-CMs do not generate translatable data without some sort of maturation. There are a wide variety of approaches used in this respect which makes it difficult for regulators to understand the subtleties of each model. This is certainly an area where a more focused approach across the industry would help.

If your compound binds to hERG and another ion channel that is expressed at lower levels in this system versus primary cells will the cardiomyocytes not be as representative as primary cells and be misleading?

The expression of several of the key channels involves in the generation and maintenance of the action potential in hiPSC-CMs is the same as found in human primary cardiomyocytes. However, there are differences in the expression with other channels. For the compounds analysed in our translational study the data from the stem cell is extremely predictive of the clinical effect. However, it is possible that if a compound does interact with a target that is differentially expressed that there may be a difference between the systems.

Considering the “predictivity” of hiPSC cardiomyocytes from Steve’s presentation and the CiPA28 study, why wasn’t hiPSC CMC assays made compulsory along with hERG and in vivo models in the recent Q&As?

My personal feeling is that there has not been enough work done yet. The data from the CiPA28 study only used 4 concentrations so as far as I was concerned there was never enough granularity in the data. Also, the fact that they compared to TdP was always an issue for me. The biomarker used in the clinic is QTc so the studies should have focused on that. I think then there might have been more confidence. Its clear from the recent HESI meeting that the FDA are not currently using the stem cell data for TQT waiver decisions, however I would argue that is because most assays don’t have a really robust translational data set that has been performed in alignment with the S7B guidance. I think this assay does tick those boxes.

For the iPSC-CM predictive QTc model, is there a specific assumed mechanistic context under which it can be used (e.g., you need to know there is hERG inhibition)? For example, was the model validated with NaV1.5 or CaV1.2 channel activators?

From the profile that you get you can make some pretty reliable interpretations of what may be going on in the absence of ion channel data, however having the additional data is always helpful since the assay can also highlight effects that are not necessarily ion channel mediated (including general tox – especially with longer exposures). The assay was validated using compound with hERG, Cav1.2 and/or Nav1.5 activity (in general blockers rather than activators). There was a wide variety of pharmacological profiles used in the initial translational study. Each ion channel or mixed effect has a specific profile that acts like a fingerprint. It’s a really useful assay when you are trying to work out what the consequence of the combination of various activities might be in a clinical setting.

What do you do for risk assessment where compounds that induce complete cessation of beating (i.e., flat BeRST signal)? How is signal amplitude used?

A cessation of beating is obviously a concern in itself and we have seen that this translates pretty week at least into rodent species (from experience). It is what we see at test concentrations just prior to cessation that is important. If you see a large increase in rise time then this suggests to cessation is Nav1.5 dependent. If you see and increase in beat rate prior then that suggests Cav1.2 block. So, the data provide a good idea with regards to a starting point when trying to derisk an adverse finding.

Did you correct the APD changes for the beating rate of the iPSC-CMs?

Yes, we did. We used the Yamamoto correction.

For multi-ion channel blockers, the PR interval and J-Tp interval on the ECG are crucial for assessing the inward current block (i.e., late INa or ICa). However, field potential recordings in iPSC-CMs lack P waves and T waves. How can the effects of a multi-ion channel blocker on field potential be assessed in iPSC-CMs?

We are measuring action potentials, not field potentials, although similarly we are not looking at an ECG response. However, for compounds with multi-ion channel block we do see specific phenotypes and the assay give us a good idea of the main effects. For example, hERG prolongs APD90, Cav1.2 block shortens APD90 but also increases beat rate (a quirk of the stem cell), Nav1.5 shortens rise time. Both Cav1.2 and Nav1.5 will eventually produce quiescence but only after you see the above effects.