Hypothesizing on how Nitrosamine Drug Substance Related Impurities (NDSRI's) are being formed?

Foreign.

My name is Ron Najaski. I'm CEO of Emory Pharma. Welcome to another Emory Pharma speaker series. Today we are honored to have Dr. Paul Harman. Paul and I basically spent two years at university of Pittsburgh. Different time. I was in the late 70s and he was in the early 80s where he got his PhD. He did his postdoc at NIH. After a couple years spending a time at Liposome Company, he joined Merck & Co. And he spent 22 plus years at Merck. I don't know anybody that spent that many years at any company. So good job, Paul. And now he's consulting and he's one of our consultants at Emory Pharma. Where we got in touch was around the generation of nitrosamine in different drugs that have really come up in the last few years. Obviously I recently presented on nitrosamine and what happens the risk analysis around nitrosamine. And we're very keen to learn more about nitrosamines from Paul. So if you're interested in nitrosamines and how they are generated as a process impurity, as different ways during the stability, this is a place to be. So thank you for joining us with that. I'm going to let Paul present. Thank you, Paul.

Thanks, Ron, for giving me the opportunity to share with Emory Pharma these two papers, you can see their titles down here that I was able to publish last year. I should point out that while I was with Merck for those 22 years, I never thought about for one second in any way, shape or form a nitrosamine in the current context at all, and had no experience at Merck at all with nitrosamines. Rather, there I did stay deep in the lab, never stepped out of the lab and published as many papers as I could around oxidative degradation, around how excipients could induce all kinds of reactions in our drug substance and amorphous solid dispersions, all kinds of things which were really much, much fun. In 2016, I retired and it was only a number of years after that retirement that I ran into nitrosamines in the context that you all know. And in seeing that topic come, I felt that at one point, not too long ago, I could make these contributions right here as you're reading here. And I should point out that at the time I was not capable, didn't have any lab access myself. So none of these measurements are my own. And this is all just kind of sleuth work, detective work of looking at what was unexplained. This is all in drug product, I should say, but. Or in the case of Renadine, it actually is a drug substance, but not the manufacturing process. This is just all drug product on stability is what I'm looking at in these papers. With that, I think it's best to start out as I'm going through here with the formaldehyde paper seems to be the one that would make the most general sense to talk about first. And in starting with that, I really need to highlight that I should point out that some of the practitioners in, in the field found this paper. I think unsettling is a word that is fair and I think that's a very reasonable reaction. And when I thought about how to frame this work, I thought, you know, if I can't explain why anyone would be unsettled by this, I'm not, I'm not giving it a very good context. So as I developed these, these first couple slides, I was intent to ultimately come across with the idea of why some folks found this work a little unsettling. So the first three slides are trying to get at that and are rather worry wordy. So I apologize for that. As you all know, in the first couple years of the nitrosamine era, it was all about small things like NDMA and drug substance manufacturing processes. But in the last several years the concern has moved into, you know, drug production, manufacturer and storage. What are the analogous things that we can take from there in terms of nitrosamine formation? So it's become really about this term ndsri, which you see defined here. Nitrosoamine drug substance related impurities. And the one that is drawing the most attention naturally is like, hey, I have my drug molecule and it has a secondary amine and I nitrosate it. Well, that's what we should be worried about in this context. So that's where the concern has largely focused. And in the Last couple years, 20, 20, 2023 was a difficult time. There was a number of voluntary recalls due to these detected NDSRIs, but there was a lot of confusion about what kind of levels are really, you know, should be allowable. Should it be more like these NDMA kind of levels or should they be larger? Higher levels? Given that this is such a larger molecule, the relative mutagenicity is not really understood. So that was kind of why these were voluntary. No one really knew what to do. This situation improved considerably third quarter of last year when the FDA released allowable limits based on a potency categorization process. They looked at all the secondary amine products that were out there, were a lot of them. They found 263 of them and said, well, we'll put the nitrosamine group on there on the secondary amine and we'll figure out what kind of limits are, you know, we're going to put on that in this potency categorization process. And that's what they did. Where are we now? Well, it leaves us where we know about nitrite ion, where it's coming from. So we know all about the nitrite ion coming in from the excipient. So that's good. There's extensive databases on this now. And generally it's accepted that more or less general tablet is going to have about 1 ppm nitrate per tablet. So at this time, the innovators of those 263 marketed products are expected to carry out risk assessments as to whether or not these are actually going to form in their products. And if a risk is identified, this is right out of the guidance. The innovator is expected to carry out confirmatory testing with an appropriately sensitive method. And there's a time limit. They're supposed to get that done more or less through all of their products by late 2025. So it's as if they're going to be doing risk assessments on the ones that they. Well, the risk assessments that indicate the most risk, they want them to test early, earlier and actually do the confirmatory test. And then as they get toward the end, that'll be the ones that are less risk. And I assume they'll still be doing confirmatory tests. Not exactly sure, but this is a very sensitive area, this risk assessment. It's impacting marketed products with limits that are already now established by the fda. So that is a rough scenario. So again, pushing that risk assessment idea a little bit more so that we can see ultimately where this work leads is, you know, you start doing the risk assessment, it becomes obvious that there's the secondary amine and the nitride ion cannot actually react. They're not directly reactive. And so something in the drug product matrix must convert nitrite to a reactive species. Oh, I meant to ask, is this cursor? This is visible to you guys? Correct. So something in the drug product matrix has got to convert the nitrite. So question is, well, how? And this first bullet's highlighting that the only way that that's ever been talked about in the context of drug product is the so called acidic pathway, which is just the nitrosation of amines and solutions of nitrous acid and that is nitric nitrite ion plus a proton is nitrous acid. So we're just talking about low PH of the solution can drive that nitrogen. So current risk assessments really only search for acidic causalities if they're looking into risk given that basically nitrite is going to be there at 1 ppm. So this work that I'm going to tell you about strives to raise the awareness that look, 50 years ago formaldehyde was found to be able to catalyze this, this reaction at neutral ph. And coupled with some of the. Well, three of the 10 most commonly used excipients are known to release low levels of formaldehyde through oxidation of their matrix. And further, the paper that I am now highlighting implies that the formaldehyde catalyzed nitrosystation is operating right now in solid dosha form and accounts for several of the six drug product recalls due to high NDSRI values and that all risk assessments need to include this new pathway. Given that a lot of folks are in the middle of those discussions with the FDA and have a lot of crunch to get through 2025, a brand new thing to consider like this that is could be very significant I think is giving rise to. It's kind of upsetting at this point to have that kind of knowledge or that kind of possibility thrown into these risk assessments. So that's where that upsetting reaction I think is, is coming from. So now we can get into the basic science of, of this paper and this is the outline. Talk briefly about nitrite. Not going to talk a lot about that. Review this acidic pathway as it seems to apply to tablets, then talk about the formaldehyde pathway, then examine these six drug product voluntary recalls. So nitride ion levels and excipient. This is just from the huge database that has been developed right out of their publication. This is eight of the top ten most common excipients. Here are the mean nitrite levels. Yes, there's a lot of variability that didn't include the ranges here. But the point is they, you know, you calculate the typical tablet composition with each one of these used at its typical weight, et cetera. And It's a general 1 part per million nitrite in a typical tablet composition. So this is kind of busy slide. I want you to focus down here first. What I really want to do is look at these new FDA NDSRI limits and think about how much of the nitrite in this one part per million is actually needed to reach those limits to sort of relate those two in our, in our head. So the FDA new limits that they came up with are only in four buckets. Has to do with the structure, the local structure and then the secondary structure. There's a whole procedure that they go through to get these values and they lay it all out there. And anyone now can take their new secondary mean drug substance and look at these guidances and put yourself in one of these categories where you're first going to be. So the, the limits are 1500 nanograms per day, 400, 126 and the percent of the 263 which fall into these categories. I just walked all through the 263 and tabulated the percents. Well, fortunately, about half of them are at this high 1500 nanograms per day. So that's good. But about the other half full 1/3, 1, 1/3, 1/3 at 400126 nanograms per day. So this column over here just comes from, you know, calculating. Well, use a 300 meg tablet that's going to have 300 nanograms of nitrate at 1 part per million. That just works through. If you take a 300 gram per mole drug, you know that you're going to end up with, okay, 100 reaction is going to be 2000 nanograms. And all I'm doing here is taking, okay, 1500 divided by 2000 75%. So for the 1500 nanogram per day, you're going to need to react about 75% of all the, the 1 ppm nitrite to get to that limit. That's kind of hard to believe that you're going to be able to do that. That's why this is a good number. If you fall here, you're in good shape. The problem starts to become half the time you're going to be, you're only needing a lot less than 75 here. 20%, 5% and 1% of the nitrite that is actually there in a 1 ppm tablet. So there's plenty to do, you know, to get you to these limits, especially these 50% of these, the time, these lower three limits, we should note that they have gone ahead and they've further looked at five of five compounds have actually been present enough that they've now gotten specific, even more specific than these. They've done the real assignment for five more and they've all come out at around 100 nanograms per day. Or less. So the ones they're really specifying even further are actually down in this lower limit. So the next slide I want to look at how these relate to the ICH Q3B 0.1% reporting thresholds. This is all from the last slide here, all we're adding here as well. Okay, so you have a 50 milligrams dose per day and 0.1%, that's your ICH Q3B, that's 50 micrograms per day. Then we're looking at how much, you know, these are compared to that. And so 1500 nanograms a day is 1.5 micrograms. So 50. So this is only third. This is 30 fold less for this big, this big higher upper bucket. So you're still 30 fold less in Q3B, not too bad. But you can see how this drops off to 125, 502 thousand fold. So roughly 50% of the time you're going to be at 125 fold or lower than QCH3.3B 0.1% reporting threshold. So these are, you know, significant challenges. And they're low enough that if you don't have a secondary amine in your drug substance, say you only have a tertiary mean, but you have a secondary mean, degraded, you might still have to pay attention. I mean, a degraded now reporting at 0.2%, that's the ID threshold. 0.2% of this is 100 micrograms per day. If you're only converting 2% of that, that's 2,000 nanograms per day of your NDSRI. So you're all the way back up here to the top of this limit. So you may not be able to just ignore impurities. As a matter of fact, in the last three years, I personally know of at least two cases they weren't recalls, but they were real time iterations with the FDA that were both impurities were getting nitroted and running a foul of. I'm not exactly sure it was these limits. It was probably the original limits before these were set. So I can't say exactly where they were, but still enough that impurities were being considered. So at this point I'm trying to get our minds back, or down, I should say, into where these kind of reactions might be taking place. And this is just some mind experiments here, wanted to get people there. It's, you know, in solid Dosha's form, all these particles are thought to have absorbed micro layers of water. And these micro layers are basically where these reactions happen. So down here I'm just trying to get at how you can get at that pretty easily. So here's a 50 micron particle. Even when has 0.5% absorbed water, you put it, you know, put the volume on there and it's wow, it only has to be 0.04 microns to hold that 0.5% weight. But that's 40 nanometers. Us older folks think of molecules in terms of angstroms, so that's 400 angstroms. So gee, if you lay five benzene side to side, that's only 35 angstroms. So if you string out this entire width of this slide, that's still only 215, 250 Angs in benzene space, if you will. And that's not even to 400 yet. So then you think about, well, Tablet has about 500,000 particles on average depending on some size that you want to use. As you know, an average of the API and your excipients. 500,000 particles like this crushed together by compaction, roller, compaction, tablet compression, whatever you want. You end up with a torturous diffusional network of drug and excipient monolayers. If you want to think about that. 500, 500,000 particles. How about a school bus? No seats, no students. That's about 500,000 golf balls. So if you're struggling for thinking about 500,000 things crushed together, a busload of golf balls kind of gets you there. So it's in this crazy torturous diffusional network that the drug is going to encounter protons, it's going to encounter formaldehyde, it's going to encounter nitrite. And this is where we talk about the excipient influences on the micro layer ph. Absolutely. So protons can diffuse and this is where we start talking about, or when they do the acidic risk analysis, you start Talking about slurry pHs, among other things. What kind of ph values can excipients drive those layers to? And these are very common. 5 to 20% by weight. Just throw the thing into, into water and you measure the ph. It couldn't really be very simpler. And the data makes sense. The vast majority of these are ph6 to 7. Sure. Cross carmelized sodium is a little lower. This is typically used at like 5%. Mannitol is actually low 5 and used when it is used at significant amounts. So a mannitol tablet could have a pretty low ph for example, but the rest are pretty high, commonly high. What about drug substance particles? They can Change ph too. Particularly, you know, we're doing secondary amines. So we have salt. So salts are often in the form of, you know, H plus a minus for secondary means. And I'm not going to get into gory detail here, but depending on the nature of a minus and the basicity of the secondary mean, there's easily calculated values to the values reason calculated to show what the ph is going to be driven to in a saturated solution and you know, reasonable number of times the most acidic component of the entire formulation is the drug substance particle. Again that can be easily calculated and the drug substance salt form is an important part of that risk profile. Now I want to talk about the traditional nitrosating agent in acid that has been talked about for many, many years. And I want to kind of trim it up a little bit as we might apply it to tablets as we've gone through to get to this point. So we're just talking about nitrite takes a proton goes to nitrous acid. That's PKA of 3.15. To go beyond this and generate these nitrous hating agents which are typically talked about dinitrogen trioxide, nitrous sodium nitrous acidium ion and nitrosyl chloride, you have to have more protons and or more nitric acid. So the first one I want to talk about is this one dinitrogen trioxide. The points here is going to be not in tablets and that's because we just don't have nearly concentrated enough nitric acid. I mean if you think about that 1 ppm nitrite and you think about all those diffusion micro layers around particles, you use estimates for volumes and such like that. You know, at best 1 ppm nitrates in, in that whole matrix is you're, you're talking 0.1 to 0.1 millimolar nitrite in the micro layers. As a ballpark estimate, that is way too dilute to even this top formation of this dinitrogen trioxide. So we're going to cut that out not for tablets. Similarly you have to get nitrous acid here and very concentrated, you know, very low ph, very high hydrogen ion concentration to get this nitrosonian greater than 4 molar. I mean we can't even get to 5 a ph of 5, 10 to the minus 5 molar. So this is really not happening in, in tablets in any kind of practical way in tablets. What's happening is one of these bottom two which still needs another prot on the nitrous acid to get this nitrous acidium ion. And in the presence of chloride, this can actually form nitrosyl chloride. I have chloride there because we got a reasonable number of HCl salt forms that are used. And so you can't discount the presence of chlorine in an HCl salt, for example. So these two are really where what's happening for a acidic tablet nitrocen agent. And this is just showing that reaction because I kind of forgot to show. Yeah, once you get here, a second can just react there. Kick off the water now and get your nitrosamine. This is how they've become activated. If you put that all together, they now have models out there for the ph dependence of nitrous acidium ion. And roughly speaking, this is log nitrostation rate at some concentration of some secondary amine. I think it's dibutylamine or something in this case. Point is that 1 millimolar nitrite. This gives you the ph dependence. It's dropping off here because you need that. You need to be protonating nitrous acid, which has a PKA all the way back up here. So this drops sharply at higher ph. And this is now this little box, this is our tablet ph. This is why we do these risk assessments around acidic excipients. Because, my goodness, if you're up here, you have somewhere in your tablet like a ph or whatever. This is like five, five there. Well, you're almost 30 fold higher in rate than you're going to be if your tablet's out here at ph 6.5. That's a lot. You're going to form a lot less NDSRI out here than. I'm sorry, Yeah, a lot less here than you are up here. So that's, that's the whole point. This is just a summary of the SITIC pathway. I don't think we need to go through it, except up until this point, if anytime we see these NDSRIs, we say, well, we must have been accessing low PH enough somewhere because we formed it. And that's all we've ever been because there's no other way. Now we move into this other way. So this is just. Now I'm going to talk about the formaldehyde pathway. This was in 1973. Again, this is in solution. We're going to start down here, start at A, and go around this way. So you start with formaldehyde, you got a secondary amine. This lone pair attacks here. You all know you get this aminal product here that can dehydrate here to form the iminium ion. And this is the reactive species. But note the iminium ions also back reacts with water significantly. Ultimately these yields in water are only 5% or so because of this back reaction. We'll come back to that later. But at any rate we have this minium ion. This is the reactive species. This N plus here is 9 for this, this bond here to be kicked back so it can relieve that charge. So this CH2 wants to be attacked by a nucleophile even as bad as this nitrite, O minus nucleophile. And that's what happens. So I'm trying to show this approach over here of these two and you know, lone pair that, that electron pair comes down, kicks it back and you get this nitrite iminium adduct. Now this would be strange enough, but the really strange part that no one I'm sure predicted, but they just discovered was that this adduct here collapses pretty rapidly. And I show the bonds rearranging like this. I don't think anyone quite knows. But the point is a new NN bond is formed. At the same time this, one of these other bonds collapses and formaldehyde is kicked out. And this species, you know, the nitrous stated secondary amine drug is actually generated. And you might think this, this, this has got to be. The efficiency of this has got to be bad. No, the bad efficiency of this whole reaction is this back reaction. When they make this, they made this back and they put it in pure very high acetonitrile with the nitrite, it collapses into this and collapses this out at like 95 yield in like 3 minutes at room temperature. So these things love each other. Once they find each other and crash down to this and expel formaldehyde, it's kind of, kind of amazing because. Well, I don't know. That's, that's, that's the discovery of science. And the ph plot is actually quite flat. There's little ph dependence from where they studied it. That's 6.4 to 11. Now we talk a little bit about. Well, that was, you know, added formaldehyde. Where's formaldehyde coming from? From our excipients. But this has been known for quite a while that certain excipients. This is that same list of top 10 excipients. There's three excipients in this top 10 list that evolved formaldehyde, HPMC, povidone and cross povidone. That's because they have groups in them. I Didn't list them all here, but they have groups like this that are oxidizable by peroxyl radical. So this is natural auto oxidation. And for these excipients they slowly give off formaldehyde as part of that oxidative process. Here I include a few others which are known to be even worse. And these are typically not in tablets but they are occasionally like ethyl cellulose and hpc. Again they have these same kind of groups and they evolve continual formaldehyde. So again formaldehyde at like Pete will come in at you know like PPM levels. If you're using some of these top these big three formaldehyde producers if you will. And again that's going to lead to something like 1 millimolar formaldehyde in tablet microlayers. Same as the nitrite, you know, rough concentration. This slide's trying to get out. If you look at that keeper and roll roller data most well all their data was really with formaldehyde, except there was one data point that was actually without formaldehyde. That was not a zero measurement, meaning they actually measured enough to put into the paper as a non zero number. And that was at ph64. If you look at that data, that data implies that at pH64 with nitrite equaling formaldehyde concentration equaling 1 millimolar, that the nitrosation yield from the formaldehyde should be almost 20 times higher than, than the so called acidic path at ph 64 which is not really acidic, it's just it has now lowered so much that formaldehyde is that much higher at the same concentration. So what that looks like if you overlay that data onto this plot, which you know, I've done here conceptually this is the previous plot I already showed you. This is the ph64 data point from 1973 sitting here at ph64.and 20 times higher than that was the formaldehyde ph independent yield at 1 millimolar formaldehyde. And this is 1 millimolar nitrite. So this, you know, is what really helped me believe. Oh man. You know, if you got like trace formaldehyde around, this is a problem when you're out here at this high PH in your tablet because now your yields are way up here instead of down on this line. But this is why we're doing the risk assessments that we didn't know that this might be here, this ph independent line here. That was what precipitated my concern. This slide here is just kind of a bridge between, well that you know, all that work I talked about as in solution, you're talking about the solid state. So here I just am trying to say, well, there's one thing going for you in a tablet, unfortunately going for you, and that is that you've done a lot of work to reduce the water content and the water activity to as low as possible. So basically you're, you're helping shut down that back reaction. So that's actually going to help some. The other is that, you know, formaldehyde is a gas at room temperature. It's very highly mobile now. So not only can it diffuse to those layers, but it gets out and it spreads around. It's, it's got extremely high mobility. These two bullets here. Trying to say that there is other papers published out there like this Waterman at al here where, where you can look at that and say, wow, you know, he had 125,000 nanograms of this reactive D imin ion per tablet must have formed and going back up here because he had a dagger which clearly had to go through this. It was not a reaction with nitrite, but something else. And the point is he must have made at least 125,000 nanograms of that because he had formaldehyde producing excipients. This was not on purpose, of course, and we only need to react 400 nanograms of something like that drug aminium complex to get into our, to get into trouble, if you will, with 50% of our NDSRI cases. That was that slide. 7. 400 nanograms is only 0.3% of the 125,000 nanograms that he happened to make under his stability conditions. Again, not intentionally. This was just exploring the mechanism and it went through that iminium. All right, so now we're down to these six cases of drug product recalls we're going to go through pretty quickly. They're listed here in the kind of order in which they actually happen from mid-2020 to 2023. And we'll take a look at. Is there any correlations with any formaldehyde generating excipients in the formulation? What can we say about these? So the first one is metformin, and this is actually an easy case only in the sense that it is a comparative case. I should point out it was actually NDMA in the case of metformin. But it doesn't, it doesn't matter what, you know, if this is actually an NDSRI or a specific one. The point is here that the problem with the Nitrocation was only for the extended release formulation but not for the immediate release. And the extended release and the immediate release used the same drug form had all the same excipients except the immediate release because it was an extended release had a lot of HPMC was used for that tablet. And as I've highlighted, HPMC is a known formaldehyde generating recipient. The HPMC itself is not acidic. It had a, it has a slurry ph of like 6. 5. It's not unusually high in nitrite. I think it was 0.7 or something. So it's the normal contributor to this one. And so, you know, such that the er, tablet nitrite shouldn't have been any wildly different. It all comes down to, you know, the differential nitrocation in that case would, would appear to be formaldehyde generating hpmc. So now you, you take the second case. So the second case is listed here. This is the ndsri. There was no formaldehyde generating recipients used. Well that's good. And the drug, if you do the calculation, the drug will not acidify the microlayers below pH6 given its properties. But Atab, this is not one of the top 10 that I listed. ATAB here does have an acid known to have an acidic slurry ph of 5.2. So if you had to look at this, you'd say, oh well, makes sense. ATAB is driving the classic pathway of, you know, acid catalyzed pathway to nitrosystation. Good. I guess these titles are kind of blocked here. This is propranolol and it's the same thing. In other words, it's one of these extended release cases is showing the problem. The immediate release is not. They have all the same excipients and extended. Except a large amount of HPMC used for the extended release tablet. And again it's all the same comments about HPMC again suggesting differential nitrication from the formaldehyde generating HPMC quinapril. Again this is dysnitrecation of the secondary amine and it has cross pobidone, one of the three. None of the excipients are particularly acidic at all. But if you do the calculations on how acidic this drug will drive its microlayer, it will, it will drive it low and it'll drive it to something like PH5. So it's certainly possible that the PH5 is really driving the nitrous station in this case, but cross povidone is probably driving it as well. So both of these routes are active in quinapril would make the most sense to me. And here's another ofenadrin. This is an extended release as well. Same thing again. Well, there's no immediate release here, but there's ethyl cellulose. It's the same as HPMC in the sense that it is known to generate significant formaldehyde. Povidone, same thing. Lactose, magnesium, stearate are not acidic. The drug will not acidify based on its properties at all. It'll only be around pH7. So this is another case. Seems pretty certain to me that it's the formaldehyde generating ethyl cellulose that are responsible. The tablet ph is actually quite high. The last case, this probably never be all put together with what we know now, but this is a case where everything's against it. There's hpmc, hpc, the capsule shell is hpmc. They also added an acidulent tartaric acid because none of this was known at the time. And the mezzalate pka, this is, this is mesalate is very low. It's essentially all, you know, has no proton affinity. And if you do the calculations with the PKA and the mesolate, it's, it's going to easily reach pH4.9. It's got everything, it has low ph for a number of different reasons. And it has these formaldehyde generators so, you know, you can't. They're all contributing. This is breaking news. This data was not published when I wrote the papers. It only came to my attention in the last month. This is a paper in print where this is now real data. I must say that the authors admitted that they were somewhat inspired by, probably because of the unsettling feeling that, that this paper gave them, that they looked at this, they measured this. Now this is in this particular frame. I'm capturing the same 1 millimolar log nitro station rate due to the usual process. And here now is the formaldehyde path at 1 millimolar as well. And sure enough, they cross down here at 5.8. They're actually, you know, similar. But that means up here at ph64 where we're trying to have our, our tablets at higher ph. Yeah, this is the three failures that I'm sitting up here because this formaldehyde line is here. If this formaldehyde line is not here, these would have been down here. So this is the, to me, the proof that in solution at a similar dilute level, here it is, the formaldehyde can compete just like this. Sure. If you're sitting down here at ph 58 on five. Five, whatever. I guess this would be. Yeah, this is like ph five. You will be dominated by the acid pathway. But the problem is, is that as we try and move our tablet ph high because of this, if we have formaldehyde here, you're setting yourself up for failure in that. Again, these are failures with the old numbers, not, not the new FDA limits. I don't know if they would have failed here. But the point is, you can see that this is very indicative of what I've been trying to say. So the conclusions for this first paper are, yeah, three of these six cases look like this formaldehyde generation route to nitrostation might make sense. Unfortunately, I believe it's absolutely operating in our solid doses forms. I don't think we're going to have any problem if everyone's willing to say, oh, well, we're going to change salt form, we're going to get rid of excipients. If everyone would easily do that, we're not going to have any problems. I think the real problems are going to be that, you know, people are going to say, well, I don't really want to give up this particular salt form and I don't want to give up Povadone, so let's see if we can manage this. So, and you might be able to, you know, given the new limits, but you'll have to go in, you'll have to have a lot of analytical work. You have to have a lot of strategic thinking and monitoring as you try and work through and say, yeah, can these be kept? Given these extra risks, I doubt if everyone's going to dispose of all the risks, risks up front and say, we don't even want to go there. But this is the choice that we now have that innovators now have to, to make. And I don't think we're going to be having a lot of recalls like we, like we did in that earlier time. I think the FDA has given us limits to work with and I think we now know maybe how to formulate and we're now aware of some more things, but it is going to take some work. So anyway, that's the, that's the first paper.

Thank you. Thank you, Bo. This is very thought provoking and, you know, useful information. Any questions? William?

Anybody is using, using like scavenger or something, mixing this during tablet formation? Well, you know, that's a very interesting question. I've been musing over scavengers, but right now I can tell you that the. Well, what's the best way to say, I think most people don't believe this yet. I think the scientific community that are really accountable for these things, by that, I mean, you know, the guys that are working in industry that are writing these risk assessments, they're not, they're not quite believers yet. It's still working itself through the system, so to speak. This paper has just come out and the guys that wrote it, well, let's just put it this way. This data to me is stunningly convincing, which is right out of one of their figures. And if you read their conclusions, it's a, They're a lot less like, wow, this, this means this formaldehyde path is right than, than I am. But they do say it. But I'm, I'm just saying this is going to take a while to get this to soak into all, all throughout and, and have this be accepted that this is something that they need to worry about. But I am convinced it is and I know I have some connections with certain companies and they're now seeing this and so this is going to have to be dealt with. But to your point, I think that's a very interesting question. I would really rather not say too much about that yet, but I have begun to think through just that. But the problem is now you start thinking about adding something new to your formulation, which, you know, that adds another layer of its own risk. And you got to weigh that risk against. What do we even. What is this risk here really in terms of how the FDA derived these notes? It's a, it's a touchy thing, but that's a great thought and I think, I think at some point it may be considered, you know, is that something that we actually do like and actually put in tablets? Yeah.

Well, it's worth trying it and experimenting and potentially building a model that, you know, you, you can see whether HPMC and various other things, you know, pushing these different levers, water, nitrite, formaldehyde generation, and potentially another liver would be antioxidant maybe such as ascorbic acid, vitamin, a little bit of vitamin C. Well, you.

Got to be careful with those acidulants because you're going to turn on that acid pathway. So you really got to be careful with that. But, and there have been, there's been a couple papers out about now can we add things to get rid of nitrite, you know, to react with nitrate and that, that looks like that may be a route too, if you can find stuff that you really want to add to, to your tablet. But there that is starting to be investigated too. So I, I think you're right. There's really a lot of, going to be a lot of interesting chemistry that can be done. In the end, I think we're going to be fine. Is this, is, this is going to be a lot more work and that's what science works with.

So we have another question with the sequestration idea.

What is the, what is the kinetic.

Rate of these reactions?

And I take it with the, with the phenomenon of water and everything else, it's a surface phenomena in that respect. You know, rate reaction versus a parent formaldehyde formation. What's the availability formaldehyde? Well, I'm not sure I can answer that question in the, I don't know in any kind of deep way that, that, that you're asking. What I can say is there's tons of formaldehyde reactions in the solid state demonstrated all the time. And so I don't know how one wants to talk about how they're, how they're happening or how, how they're not. But for example, the formation of this iminium ion, this is well known to be able to form in the solid state. So if your question is what is the details around how this is actually coming into contact with a second mean? I, I don't know that I can answer that. I'm sure that in water this is in an equilibrium with its hydrated form. I'm not sure it would seem that the second mean wants this form, but they're all in equilibrium, et cetera. I don't, I don't know how to answer. I don't know that your answer that, your question has an answer that I can answer in that way. I mean, these things get through these layers and they become, they get into reaction contact with each other. And this is how all degradation occurs in the solid state. Everything we've ever seen. So I don't think this is any different. This is probably worse because one of the reagents is a, is a gas and that just helps it get through the whole thing even faster. In my view, Paul, you need water.

For this, you know, chemistry to go through.

Well, yeah, that's, yeah, sure, sure. That's what this is. That's what this is all about. Meaning this crazy thing with the micro layers. We all have water. Every, every tablet on, on the planet has, has a, is, is viewed to have some amount of this adsorbed water in. Typical KFS are 2, 3, 4% for, for a tablet. The idea is that I don't know how thick you want to view this layer. But even at half a percent conceptually, using simple hard spheres, there's tons of water to do a reaction. And even, even in Italy. Yeah, this is all this kind of micro layer stuff here. Absolutely. The ranitidine case is one that you're all very, very familiar with. So we'll, we'll go through this pretty fast. By the time 2019 rolled along, you guys were the first ones to show, wow, this API alone can actually grow levels of ndma. That was the first time that had been shown for any compound in the pure drug substance alone. So by 2020, there was no understanding of how this could happen. FDA recalled it. Since then, a number of reports have gone in pretty good detail trying to figure it out. No one has been able to suggest a chemical path and reninidine products remain off the market. I was lucky enough to come along and know a lot about oxidation processes, particularly peroxy radical. When I looked at this in the middle of this whole formaldehyde thing, the two came together and I was able to propose a detailed mechanism that made sense to me and at least the reviewers. And a simple mitigation procedure comes out of that mechanism. So that story is this talk. So I'll introduce the problem very briefly and then we'll talk about peroxy radical. It's all about peroxy radical and application and apply it to renitine and how to get DMA and then, and the nitrite that is released from rinidine, how to get them react. That's really the same question we already went through, but in this case it's the acidic pathway joins these two and a summary of the mechanism and a control strategy. So this is the question here is ranitidine, somehow previously unexplained DMA is released, if you will, from, from here and the nitrite ion is released out of this vinyl nitro moiety somehow. And then somehow the DMA and the nitrite ion get together to form ndma. This is the usual question of, well now we have an aldehyde path or an acidic path. It's going to be the acidic path here. But how to get this all done? So the relevant findings of the innovator. So they looked at the known degradation pathways. They couldn't explain it. They said, you know, the NDMA has a time and temperature dependence. Seems like it's a, like, like it's a degradation process. Right. The role of drug substance purity was. It was examined carefully and they said, no way. And the thing that was really wild is that there's two different crystal morphologies that they identified in their drug substance. And one had big issues with NDMA and the other did not. And these two morphologies absorb different amounts of water and the final isolation solvent on their drug substance surfaces. So two different morphologies, two very different results. The other thing is they had data which would suggest that the drug substance grew and NDMA even worse than the drug product, which was kind of unusual in the general sense. So the first thing they, they did was they were so flummoxed that they could not explain this chemistry, that they did a simple labeling experiment where they made, they deuterated this whole end. They put N15 on this, this is 100 label. And of course when they heat that, the NDMA has to be doubly plus seven because there's nowhere else for anything to come. So they didn't prove much there. But the point of this was that now they can get this and they put like x milligrams of that in a certain, in a certain amount of solvent and dissolved it. Then they got the same X milligrams of the unlabeled material, added that to the solvent and then co crystallized that and then put that up on stability and looked at how, what was the labeling in the ndma. And if this was really an intramolecular reaction where this was somehow bending around and having some kind of direct interaction and ending up with this. Well, that labeling experiment then would have shown that all the label would have been either +7 or all the label would have been unlabeled NDMA 50% of each if it was completely intramolecular. And that's not what they found at all. They found the label was distributed basically 25% between 7, 6, 1 and no label, which just means it was a completely unrelated release of this and a completely unrelated release of this. And then they were getting together randomly and the label was completely scrambled, but they had no idea how. So now we get the peroxy radical chemistry. So peroxy radical has two basic types of reactions. So you can look at this as I'm babbling on here. The first one that that matters that, that it does is it can abstract hydrogen atoms from very weak CH bonds bonds. When it does that, this peroxy radical abstracts that hydrogen atom and gets to this hydroperoxide, leaves a carbon centered radical here. Then oxygen adds to that and then you get a peroxy radical back and this just keeps going over and over. And the extent to which this reaction can even happen, proxy radical is very selective. So it really has to do with how energetic is this bond, which is typically about 370kcals per mole compared to this CH bond, the one that's trying to extract a hydrogen atom from. If this is real strong compared to this, it doesn't react. I mean, that makes sense. And this is weak compared to this. Wow. This reaction can happen quite easily. So that's just usually accounted for just in the bond association energy. So below 370, which is this, it will react. Above 370, it reacts slow. So if you just look at that, those words in a different way, that's just this here's bond energies all the way up. Here's peroxy radical sits right in here. So if it has bond energies above this, these things up here don't react. This is slow reaction with proxy radical. Things that are kind of similar, they're moderate. And things that are way down here that have very low CH bond energy. Antioxidants being the lowest. These are really fast here. It's actually O H bonds donating hydrogen atoms. Point is down here, things react very fast with peroxy radical. Very few things are down here, huh? So now you look at red editing. Previously it was never able to calculate this bomb, but now post the paper, there's a government sponsored VDE calculator, Alphabet, machine learning derived, fast, accurate bond and you can load the whole structure in and it calculates the bond energy to within 1 kcal. Hey, thank goodness for machine and AI derived learning, huh? So loaded this in to the program, stuck it in, and out came that CH bond energy. It's down here. It is the lowest CH bond energy I have ever seen. It's down where the things are typically antioxidants. So the point is it is very low energy and its own peroxy radical would have no trouble at all reacting with this bond 335, very, very low. So if we let a proxy radical approach the left side of ranitidine to this low energy CH bond, well, this is what happens. It goes through the usual peroxy radical thing. Peroxy radical tips, takes that hydrogen atom right out and that this becomes that. This unpaired electron sits here, adds oxygen. And now you have a ranitidine number one peroxy radical. That's the first molecule. I'm just labeling it. Okay, now this does the same thing. So this goes now to hydroperoxide because it took the it does the same thing. So you get a whole bunch of the hydroperoxide coming out and you're always got this peroxy radical itself running around. So what oxidation products are likely to accumulate longer term from. From this. This is the first level slide, slide one. Well if you look at anytime you have proxy radical you can get the disproportionation of this radical. For this case, that's the hemiaminal and. And this amide. So it's all about the hemiaminal here. So here's a reninidine. Here's the other product that renamine renanine hydroperoxide. This can decompose in a lot of ways and they typically do. This thermalizes across this bond splits homolytically. This goes to O dot. That's a very strong radical. Abstracts hydrogen atoms and anything goes to the hemiaminal. That's what you would expect it to. This can also eliminate hydroperoxide. Often happens not a bad leaving group. As soon as you get this. This can react with water reacts here and goes back to the hemiaminal. So you make the hemiaminal. And there's only one thing the hemiaminal really wants to do and that's get back into equilibrium with the. With the amine and the. And the aldehyde from which it can be viewed as coming if you will. So this reaction just goes boop boop boop off kicks dma. It is one of the natural products of peroxy radical oxidation on the left side of rinidity. The other peroxy radical path, unbelievably is addition to an olefin bond which happens to be also sitting on the right side of redinidine. There's an olefin bond. This reaction is extremely well studied and known just like the other one. So proxy radical adds to a double bond. When it does so it leaves an unpaired paired electron. Now it has a choice either oxygen adds very very fast which gets you this kind of peroxy radical. This is just oxygen, molecular oxygen adding here. But in this case this thing can also interact with this OO bond and cleave out any epoxide. So these are two well known reactions of epoxy radical. I'm sorry peroxy radical reaction with olefin. And so there we go. The same peroxy radical that comes to the right side of ranitidine and does the reactions I just showed, they all lead inevitably to the same thing, the expulsion of nitrate. You can follow any which ones here given the limit of time. And I know we barely Read in doing this particular presentation. So I probably won't go through this, but this is pretty straightforward. Peroxy, radical disproportionation, alcohol, it's, it's all the same stuff. Nitrite will be produced from the right side. So we have independent paths here and here. Producing nitrite ion whoops. And DMA just like is needed explains the labeling experiment perfectly. Now the question becomes all right, well how do we get these together in, you know, in the API to make ndma? And based on what we've known, it's either the acidic or the aldehyde. Well, there's no excipients to produce formaldehyde, but as I showed, a couple ranitidine aldehydes are produced. But if you look at the drug substance absorbed solvent content which that paper had and you calculate the relative concentrations of water, methanol and isopropanol, there's over 100 mgs per milliliterine that can get into solution in both of those different morphologies. At 100 mgs per milliliter, that is A, the pH should be something like 5. Now some error is in there because it's not 100 water but 100 mgs per mil ranitidine HCl will be acidic. It also brings in, you know, we estimate, well, we estimate its solubility is 0.3 molar in those layers. At any rate brings a lot of chloride IONA as well, of course. So we also could be enabling the nitrosyl chloride, the nocl. That's the same as the other nitrosating agent, the nitrous acidium. It's the same thing. The point is it really argues that the acidic pathway is likely how they're joining. It's also interesting that the NOCL pathway could explain the two different morphologies actually being very different in NDMA formation. I was was shocked that I was able to offer an explanation for this. But if it really is the nitrosyl chloride path, well that's got to be any. That is an equilibrium with, with water. This equilibrium constant is known and is and is published. There's reference for and one morphology has eight fold higher water in its micro layer than the other. That's just based on the numbers that they gave and how they absorb different amounts of water and solvents. So good old Le Chatelier. If I'm saying that that right, If I have 8 times more water here, I push this to this side in comparison to the other and that means less yield of the Nitro. The nitrosating agent, nitrosyl chloride. And sure enough, the morphology that has eight times higher water is the morphology that has less problem with NDMA formation. So I. I was stunned, but yet had to further believe that this is what's happening. I will lay off this bullet, given the time. And so here we are. Everything can be explained. Peroxy radical reactions can explain both releases independently from each side. The drug itself is extremely reactive with peroxy radicals by that calculation. Never seen a more reactive one. The independent peroxy radical left and right. Yep, did that. The NDMA does have a time and temperature dependence that looks like a degradation process, because it is. It is a peroxy radical oxidation process which has the typical kind of temperature and humidity dependencies. The two morphologies grow different amounts of NDMA due to different NOCLs. 8 times more water present on one gas. We're going to skip this bullet for time. A full chemical rationale appears, and the mitigation procedure is simple. We're on the second to last slide. Well, proxy radical chemistry consumes molecular oxygen at every step. Without oxygen, none of this can happen. So oxygen scavenging technology for ranitidine, in my view, will completely shut this down. That's the only thing, I think, that hasn't been tried, and that's why nothing has worked. So the kind of simple experiments that emery detailed in 2020, but instead of doing it in the presence of an oxygen scavenger. Course, you got to have a seal on this thing. It's got to be a seal, you know, to keep the oxygen to zero. I'm suggesting we'll show no increase in NDMA levels whatsoever. Of course, assuming I'm right, and I'd be delighted to help in any way possible. And I think that's it. Yeah.

Thank you, Paul.

You.

You've given us a lot of food for thought. Obviously, some of your hypothesis are worth exploring and we will explore. And for sure, though, there are opportunity for us to do some publications together in the meantime.

I would really love that. That would be really, really super.

Yeah. In the meantime, if for those of us who are listening to this talk, either through Emory Pharma's website or our YouTube channel, please feel free to contact Emery Pharma by sending us an email@mary pharma.com or contact Paul through his consulting company, get involved and help you figure out how we can minimize your nitrosamine formation and effectively figure out the parameters of your drug substance, drug product in terms of what will make it form nitrosamine and put some controls around your final product. Thank you all for listening, and we appreciate Paul's contribution to this area of research.

Thank you very much. I appreciate it.

Thank you.