E66: When Drug Candidates Miss the Mark: Off-Target Liability

Mary Parker

I'm Mary Parker and welcome to this episode of Eureka's Sounds of Science. A drug target is typically a protein in the body that's associated with the disease. The target could be a cause of the disease that a drug is trying to block, or it could be a potential treatment that the drug is trying to activate. Hitting this target is what makes the drug effective, but what if they hit more than one protein? Joining me to discuss the implications of off-target activity is Stan Spence, Senior Principal Scientific Advisor at Charles River Laboratories. Welcome, Stan.

Stan Spence 

Hi Mary, it's nice to be here.

Mary Parker 

It's great to have you here. So Stan, to get us started, can you tell us a little bit about your background and what brought you to this type of research?

Stan Spence 

I have a background in developmental and molecular biology. And I started in drug development or toxicology way back in the early 90s. And I've spent a long career in pharma and biotech before joining Charles River. And I've developed many drugs for both general medicine as well as oncology. And in doing in transitioning from academia to industry, I really focused all of my research in the area of toxicology and safety pharmacology. And that's kind of the sweet spot for my experience, is the transition from late discovery through IND enabling and beyond.

Mary Parker 

Just for my own edification, have there been any drugs that you've developed in your career or you know, maybe that didn't make it to market, but something that had a very surprising off-target effect that got the drug, you know, maybe canned before it got any further?

Stan Spence 

Well, I can think of molecules, yes. I can't speak to particulars because I'm still covered under confidentiality. But there have been drugs that I've developed that have had off-target effects that have resulted in significant liver toxicity that we terminated before we brought into the clinic. And I think in that regard, an early termination is just as valuable as bringing a molecule forward that doesn't have those liabilities. Because eventually a molecule with these types of liabilities isn't going to make it to become a drug. And so you want to, it's best to fail early, identify these liabilities early and terminate a molecule early if it's not going to become a clinical, actually become a drug.

Mary Parker

Can you tell me what is the difference between off-target binding and off-target liability? I have no idea so you're gonna have to explain it to me in small words.

Stan Spence

Yeah, so off-target binding is merely the ability of the small molecule or antibody to bind to a target other than the intended pharmacologic target. An off-target liability would be binding and activating or inhibiting an unintended pharmacologic target that results in an adverse clinical event.

Mary Parker

Okay, so basically an off target binding is it just might hit but might not do much or have any side effects. Whereas off target liability is it might have adverse side effects.

Stan Spence

That's correct. It really depends on exactly what the off target is and the potency in which it inhibits the off target relative to the principal target that you're trying to drop.

Mary Parker

Okay, that makes sense. So what stage of the drug development process focuses on off-target effects, or is it something that kind of has to be watched for at all stages?

Stan Spence

Well, I tend to think that it should be covered more at all stages, but when you're elaborating a new small molecule or a new antibody, you tend to do this during lead optimization. And then if you discern an off-target effect that wasn't picked up during lead optimization, you would probably go back and try to deconvolute, try to figure out what that off-target is so you could develop a higher throughput screening assay for your backup candidates. Sometimes you can have an off-target effect that manifests in the clinic. It's probably a worst-case scenario, in which case then you want to determine what the mechanism of that off-target effect is so that again you can come back and create a high throughput screen to iterate a better molecule, a backup molecule.

Mary Parker 

Yeah, that was going to kind of be my next off topic question is, you know, each stage of the drug development process uses different models and none of those models are perfectly emulating a human being. So I imagine that it's not terribly infrequent for it to get to the stage where it becomes, you know, active in a person and has an effect that people weren't expecting. I think we narrow, we try and avoid that as much as possible using every type of model we have available. But until it comes into the actual full human being specimen, you know, the complexities of our bodies is a lot more than can always be accounted for in the process, right?

Stan Spence 

Yeah, that's correct. Each off-target effect, there are certain animal models that are more appropriate than others. And I can give some examples along those lines, but you wanna make sure that once you identify an off-target, if you end up finding it after you're in humans, once you identify what that off-target is, you want to utilize an animal model that best recapitulates that effect that you would anticipate in human.

Stan Spence 

And I could give a few examples around this if that would be helpful. So if you take for example, seldane or terfenadine which was an antihistamine that was marketed in the 80s and 90s and pulled it off the market in the 1990s due to a lethal arrhythmia called Torsade de Pointe, that molecule actually inhibits a potassium channel called HERG or KV11.1.

Stan Spence 

And in rats, KV11.1 is not expressed in adult rats. It's only expressed embryonically in the heart of developing rat embryos. So obviously the adult rat would not be a model for assessing off-target effects associated with HERG. However, the dog and the non-human primate would. Another example is troglitazone, also known as Resulin, which is a PPAR agonist that was developed for type two diabetes. This molecule actually caused significant liver toxicity and liver failure in some patients. It was subsequently found out that this molecule inhibits a bile acid pump known as BCEP. It's called the bile salt export pump.

And when you inhibit this pump, bile salts accumulate within the hepatocyte and this causes cholestasis and they'll reach toxic levels within the hepatocyte and eventually cause the hepatocyte to die. The rat would be an inappropriate model for studying the effects of B-set because the rat has a CYP2C8 which oxidizes bile acids into more hydrophilic substrates that can be eliminated in different ways. 

Stan Spence 

I could also give a third example if you're interested. So you might remember back in the 1990s, there was a weight loss combination known as fen-phen. It consisted, it consisted of fenfluramine and phentermine. And fenfluramine, which was one of the components of fen-phen, was known to actually agonize a serotonin receptor known as 5-HT2B. And when you agonize 5-HT2B, it's expressed in the heart valves. And this results in remodeling of the heart valves and they become stiff. And you end up with valvular regurgitation during cardiac contraction within the ventricle in particular. And this can result in heart failure over time. And so some of these patients who were exposed to fen-phen actually ended up with valvular replacements, cardiac valve replacements. And the drug was subsequently removed from them, taken off the market. And so we have, in this case, the rat would be a perfectly good model because we've shown that in the presence of fen-phen or 5-HT2B agonism, you can get valvular remodeling in the rat, as well as other species, but the rat would be a good model for that.

Mary Parker 

So after events like this, where it makes it to clinical trials or even to market without these being known, has the FDA responded with new regulations to avoid those sorts of things from happening in the future?

Stan Spence 

Well, the regulatory agencies in general have responded with regards to BSEP. The EMA requires BSEP profiling in the transporter package prior to, you know, for an IND. The FDA recommends BSEP, but they don't insist upon it like the EMA does. Other health authorities have different requirements for BSEP, but in general, it's recommended to profile for BSEP.

Mary Parker 

Okay, so they do tend to course correct when things like this happen. Yeah, on the other hand, off-target effects can sometimes be accidentally beneficial, right? Can you give an example of that?

Stan Spence 

Absolutely. Yeah, that's correct. This is particularly true in the case of kinase inhibitors. It's been shown for some kinase inhibitors that drugs that inhibit multiple kinases involved in cancer, such as sutent, can be more effective than those that are highly selective for one particular target. So off-target inhibition or multiple kinase inhibition in the case of oncology might be beneficial in certain circumstances, depending on the off-targets that you hit in addition to the highly desirable ones, a multiple kinase inhibitor can in some instances be more efficacious than a very specific kinase inhibitor. It depends on the cancer that you're targeting and what its kinase profile is.

Mary Parker 

Okay, and this is not to even bring into it drugs that have been found to have effects other than their intended effects, right? For example Viagra or Ozempic or things like that. Yeah.

Stan Spence 

Yes. Yes, you bring up a good point. I think that Viagra was originally being developed for cardiac hypertension or high blood pressure, and then they found the other beneficial effects in usage for it for ED, for erectile dysfunction.

Mary Parker 

Mm-hmm. Yeah. How does choosing the right model help during drug development process with off-target liability?

Stan Spence 

Well, if you're not choosing the right model to de-risk your off-target, then you can be led to believe that a molecule has a favorable profile. And then you'd be bringing into the clinic, which is the most expensive phase of development. And if a molecule fails during clinical development, you've lost an awful lot of money and time. And then you need to go back and try to create a backup molecule that doesn't have the same liability. In addition, you need to figure out what that off-target liability is, so that you can find some high-throughput in vitro screen that you can add to your lead optimization funnel to find a molecule that doesn't do that. And this is particularly difficult with small molecules because a lot of times you're making small iterations to the structure of the lead molecule and the off-target effects are structurally related.

Mary Parker

So just out of curiosity, when we're talking about these off-target effects, are there any in-vitro models or any non-animal models that can give you hints as to off-target effects being possible?

Stan Spence 

Oh, sure. In the case of HERG, we have human embryonic kidney cells that have been engineered to express HERG. And so those in vitro profiling has been available since the late 1990s to evaluate potential effects on HERG during the lead optimization stage. And sometimes if the chemist knows there's a liability early on, they'll actually screen well before that, do some spot checking for HERG binding, and then roll in some patch clamp assays later during lead optimization or during early non-clinical development.

Mary Parker 

And also just out of curiosity, is there any way that AI could be useful for predicting some off-target effects of any new compound?

Stan Spence 

Well, there's quite a few models, SAR models, structure activity models that are out there, which can be done in silico where you're looking at the structure of the molecule and we know that certain structures are associated with certain liabilities. That can give you some idea of what to screen for early on in a functional screen. So yes, I think AI is going to have a place in the future, especially with regards to in silico analyses.

Mary Parker 

Great, cool. So what does it mean for a drug candidate to be de-risked? Like what are the potential consequences for let's say a small company if they don't do it properly?

Stan Spence 

Well, it's like I had mentioned before, if you don't appropriately de-risk with the right model early on, and then you find a liability during clinical development during let's say phase one or phase two clinical development, in the worst case scenario, you'd fail during phase two or phase three. And you've spent all this money during non-clinical and clinical development with clinical development being the most expensive. And that's the greatest liability to a small company, I would think, is failing late. Because you only have so much money, the burn rate being what it is. And so you wanna place your bets on the best molecules you can.

Mary Parker 

Okay. Are there any classes of drugs or classes of diseases that tend to have drugs that are more prone to off-target effects?

Stan Spence

It's interesting you should mention this because most marketed drugs actually have a fair amount of off-target binding and off-target pharmacology. Most marketed drugs can inhibit up to six other off-target or up to 12 other off-target proteins or enzymes or what have you. But it really depends on the margin of safety or the relative potency for those off-target effects. relative to the intended pharmacologic effect. And as a molecule becomes more lipophilic, the off-targeted effects tend to become more predominant. But again, it really depends on the magnitude and the potency of those off-target effects relative to the intended pharmacologic effect.

Mary Parker 

Right. Like you wouldn't want there to be any risk of, of a adverse off-target effect for something like a simple painkiller for headaches, but for something like a life-saving cancer cure, you're willing to put up with a larger threshold of off-target effects since, you know, the patient needs the medicine more desperately.

Stan Spence

That's absolutely correct. For a life-saving indication like oncology, you can deal with a lot more baggage in terms of off-target effects because the risk to benefit ratio is more favorable for development. But if you're developing a drug for general medicine indication that its needs are fairly well met, then the bar for safety is extremely high.

Mary Parker

Yeah, absolutely. So do you have any advice for drug developers who have a candidate or a few candidates and are looking to plan out the rest of their testing to see which one will be the best fit? What should they keep in mind when they're doing this related to off-target effects?

Stan Spence 

Well, there's a number of off-target panels, in vitro panels, that can be done to assess off-target effects that are associated with adverse clinical reactions. And I think about the Eurofin Safety 47 panel, which every company does and is actually required for IND and is placed in the secondary pharmacology section of an IND. That should be done during, probably after lead optimization or during lead optimization. And that'll give you some indication of whether or not you're going to be inhibiting pharmacological targets that are associated with adverse clinical reactions. Of course, there's HERG as well. And all of these assays are well known to most drug developers and typically deployed. I think one of the important aspect is when you run into a clinical toxicity and you're committed to a target, it's very important to figure out what the mechanism of that toxicity is, you know, reverse translate what that molecule is doing so that again you can develop some higher throughput screening assays that you can deploy during development of a backup molecule.

Mary Parker 

Okay. Okay, well thank you so much, Stan, for joining me. This has been a fascinating look into this aspect of drug development, and I appreciate you bringing your expertise to us.

Stan Spence 

Thank you, Mary. It's a pleasure to be pleasure to participate today. Thank you.