Hi there, my name is Brendan Roche and today I want talk to you about the importance of chiral crystallization strategies and how APC approaches problems in this area - in particular dealing with chiral purity and yield. So, first of all, I'll just give a brief overview of today's talk before I start. Firstly, I'll talk about the background of chirality and stereoisomers and their importance in active pharmaceutical ingredients. This will be followed by an introduction to ternary phase diagrams and why they're so important to the correct design of an anti-solvent crystallizations. After that, I'll present some case studies that we have carried out. So case study one is about ternary phase diagrams as a tool for process definition and case study two will be the application of ternary phase diagrams in diastereomeric salt isolation. For both cases I'll discuss the problems posed, how we tackle them, and then give brief conclusions.
So, first of all, just to talk about APC and our partners. We are a contract research organization based in Dublin. We employ over 200 scientists and engineers working across, a broad range of areas from small to large molecule. We collaborate with biopharmaceutical and pharmaceutical companies, across the globe, to accelerate the delivery of medicines to market. And we do this all through our data-rich process development ethos. So, at APC we offer value through the synergistic relationship of chemistry, biology and engineering departments. This allows for rapid process definition, feasibility, characterization and development studies.
So first of all, some background. Prior to discussing any of the case studies I've mentioned earlier, I'm going to do a brief recap of chirality. So, with altered chirality can come alter physiological and pharmacological effects. For example, the stereoisomers of the APIs that I’ve shown just below are beneficial treatments for nausea, arthritis, and tuberculosis. Unfortunately, their chiral analogs caused disastrous effects such as blindness, toxicity, and androgenic fetal abnormalities. So I'll just show a brief glossary of key terms that are important and that will be noted throughout this webinar. So chirality, it's a geometric property of a molecule. A chiral compound is non superimposable with its mirror image. Enantiomers are chiral compounds that are non- superimposable mirror images, diastereomer are chiral compounds that are not mirroring images of one another and are non-superimposable. Enantiomer pairs have identical physical properties, but diastereomers will have different physical properties which will come into play. When I talk about the second case study, this difference can be observed in the solubility design space and our solid phase characterization.
So when we are talking about the solid phase characterization, this is a key step in the design of an enantioselective crystallization. This can help us understand the arrangement of the crystal lattice and if it is crystallized as a racemic conglomerate or a solid solution. So a conglomerate crystal lattice arrangement is desirable. This solid phase characterization can be achieved using binary phase diagrams or ternary phase diagrams as shown below. So during the binary phase diagram characterization, we use DSC to monitor the melting points of the solid mixture of our chiral compounds. And then during the ternary phase diagram construction, we assess the solubility of solid mixture and plot this on our ternary phase diagram where the eutectic point is the highest solubility point of that mixture. Now due to enantiomers having identical physical properties, we would expect an enantiomer to have a symmetrical diagram. While if we're looking at a diastereomeric system, we would have potentially one-sided system or a non-symmetrical system. But you'll see examples of this in case study one and case study two in a moment.
So, on the bottom row here, you can just see an example of some ternary phase diagrams. But as I said, we'll discuss these in further detail when we move on to the case studies. In our first case study, I'll talk about ternary phase diagrams as a tool for process development. In the process flow diagram shown below, a hydrogenation reaction and a subsequent purification is described where we have a number of charcoal treatments, re-slurries, distillations and hot filtrations which were carried out before final isolation of the API. This would be considered a lab- scale time intensive process. One of the obvious areas for improvement on this process was time - taking an average of one week to go from reaction to isolated product. Beyond the process time, some of the key outputs were inconsistent, such as e.e., yield, solid assay and a large variability in the heavy metal purging. APC was tasked with measuring solubility and the effect that residual heavy metals may have on the conglomerate system. Solubility was collected and an anti-solvent crystallization was developed and optimized so we could isolate the desired in enantiomer. After this, the optimized process was scaled up to 10 grams and subsequently a hundred grams to demonstrate the process robustness.
So key to all projects here in APC is the “process establishment phase”. This allows us to understand the key quality attributes associated with each process. To gain a better understanding of the process, APC employs a data-rich experimentation ethos. This allows us to gain key information for each unit operation so we can assess their efficiency with regard to assay impurity, yield loss, and in this case heavy metal traces for each reaction. In charcoal treatment, the assay in chiral purity increased. When we tracked the heavy metal content throughout, it was noted that neither of the charcoal treatments were beneficial for purging these impurities. The water and IPA washes were shown to be a lot more effective in this regard. And this signalled early on in the project that charcoal treatments were ineffective and could potentially be removed from this process.
Mass balance assessment of each unit operation was carried out as part of the process assessment for this. Each process waste stream was analyzed by HPLC shown below. About 10 milligrams per gram of API is lost for each reaction that we carry out. When this is assessed by normal phase HPLC, we see that there is a 50/50 mixture of the desired and undesired in enantiomer lost to the liquor. This highlights the important learning that properly selecting your HPLC method for solubility can affect your process. And this is also a common development pitfall.
At APC we use an in-house platform for early phase solvent selection, solubility assessment and crystallization design. We leverage the use of in silico computational screening to predict their solubility and give the relevant solvent propensity and impurity purging. This allows us to screen a variety of solvent classes and carry forward a shortlist of desirable solvents for experimental validation through two point solubility. From there, small scale crystallizations will be carried out. By doing this, we can maximize the amount of information that we gain on the potential solvent system. So, shown below is data from one such computational screen. So, 27 potential solvents were assessed. We rank them by relative solubility of our desired in enantiomeric compound. This helps us identify good solvents. In this case you can see DMF and DMSO are potential anti-solvents. The graph on the right hand side shows solvents with a high propensity to form a solid it and this is one of the tools we use to help us narrow down the list of potential solvents. For example, we look here you can see THF and formic acid removed from any subsequent experimental screen due to their solvate-forming propensity.
So to build a ternary phase diagram, we generally conduct solubility studies. Two common methods employed at APC are isothermal solubility and Crystal 16. The isothermal method is generally paired with HPLC sampling. To carry out the isothermal solubility, we need to synthesize the opposite enantiomer of the API that we desire. In this case, we did it by modifying the system we used to measure solubility. We add an excess of solids to a known volume of our chosen solvent. We allow this to equilibrate and sample the liquor by chiral HPLC and we calculate the concentration in milligrams per gram of each enantiomer in the liquid. We then convert this to weight percent and plot it. On our ternary phase diagram you can see a representative ternary phase diagram in the bottom right hand corner. So as this is an enantiomeric system as we predict, we would see a simple eutectic point at a 50/50 mixture.
In the second case study, I'll present a system which will show an unsymmetrical ternary phase diagram to verify a simple eutectic system. We can carry out one simple experiment. Saturated samples of a desired and undesired and enantiomers are individually prepared and their isothermal solubilities is checked by HPLC. From there we add an excess of the opposite enantiomer to each liquor and allow it to equilibrate and sample once again by HPLC after equilibration. In isothermal conditions, a simple eutectic system should converge at the same eutectic point regardless of the starting enantiomer, provided if both are in excess. This was the case shown in figure B confirming that a simple eutectic system was applied. At this point in our development journey, we possessed a keen understanding of the solubility of our material and how to construct the solubility design space.
Another key objective is to understand if the residual heavy metal impurities affected the solubility of the API. To do this, we isolated solids throughout the process and assess them by chiral HPLC. Solubility of the pure materials and the crude materials at the midpoint had a similar solubility. The crude reaction output solubility was also collected again, even with high levels of rhodium, zinc and iron. It possessed a similar level of solubility. While it was theorized that heavy metal content could have a a very large impact, it highlights the importance of collecting efficient solubility data.
Using our ternary phase diagrams, a potential process roadmap can be constructed. As a proof of concept, a crystallization is executed starting with 70 to 30 weight percent desired to undesired enantiomer in a 50/50 IPA/water mixture. As with all process development in APC, we utilize inline PAT with our data-rich experimentation approach. In this scenario we utilized FTIR and FBRM to track our seeded crystallization. After effective seeding and dissolution, we observed desaturation as indicated by figure 4. During our cooling to our isolation temperature, we observe a secondary nucleation event by FBRM and FTIR. We also tracked this experiment by HPLC. Here we can see an initial desaturation and that yields are desired enantiomer with the undesired enantiomer remaining largely in solution. We do, however, see secondary nucleation at 30 degrees and we see desaturation of both enantiomers on our ternary phase diagram. We can track our process from seeding to desaturation to secondary manipulation. We cross the phase boundary where we build supersaturation of our undesired enantiomer and to our isolation point and our eutectic composition.
Our first case study, we employed a ternary phase diagram as a key tool in the development of a chiral crystallization. It enabled the development of scalable, shorter and more robust process where the unnecessary charcoal treatments, drying and distillation steps were removed. Effectively, the resulting process gave us an improved yield and enantiomeric purity assay, and better heavy metal purging. In our second case study, I'll once again show the construction of a ternary phase diagram while we show how they can be used to isolate diastereomeric salts and how the system could be used to construct a semi-continuous process. Diastereomeric salt is synthesized by reacting racemic, acidic or basic compound with an acidic or basic resolving agent. These can be separated by fractional crystallization and once isolated, the salts can be neutralized to then give the pure in enantiomer as highlighted at the start of the webinar earlier. And in this case diastereomer salts tend to have interesting properties which can be leveraged when designing a crystallization. They can have high differences. So your choice of solvent and resolving agents can be key. The morphology of the diastereomer crystals can be distinct. Also, they can have varied thermal stability and a crystallization behaviour.
Lansoprazole is one such example of an API and has used diastereomeric salt isolation during their development. We are talking about ibuprofen here. The diastereomeric salt in this case will be referred to as the end salt. First we characterized our salts by HPLC and XRD and DSC then in an analogous manner. In the first case study we conducted thermodynamic solubility of the pure diastereomeric salts by HPLC. Following that, we conducted solubility of the solid mixture. Then we use Crystal 16. And this allows us to define the meta stable zone width of our salts, which is highly important when designing our crystallization. The collected, solubilities of the salts were presented as weight percent in this case and we add them to our ternary phase diagram, which is shown in the bottom right hand corner. So as you can see, the eutectic point is observed at 34 to 66 weight percent of the sample.
If you look at the left, you can see the solubility points of each salt and that allows us to define the metastable zone width in this case. As you can see, the P salt has a much higher solubility and a greater temperature dependence with its solubility. This is indicative that a cooling system for our crystallization would suit for the isolation of the P salt. Below, on the ternary phase diagram, is the method we have devised for isolating both of the diastereomer salts. If you start at 0.1, we have an equal mixture of the P- salt and the end salt at 30 degrees Celsius. This feed is then added to an unseeded isothermal vessel, which crystallizes our end salt in high purity leaving us with a P-salt enriched liquor. This liquor is then seeded with P-salt and a cooling crystallization is conducted to 20 degrees Celsius. This allows us to isolate high purity P-salt. The resulting motor is slightly enriched with the P salt, so it's mixed with an N enriched feed at 0.4, which allows us to return to our starting composition. So to conclude our second case study, we have shown how we can build a ternary phase diagram and use our solubility studies to facilitate the separation of diastereomeric salts in high purity and efficiency.
A simple eutectic point at 34 to 66 weight percent and P-salt was observed. This is something which is characteristic of systems like this. Also of note is the schematic on the right hand side, which shows how amenable this process would be to a semi continuous system. If you'd like more information on this specific case study, please have a look at the paper below. And finally, I'd just like to say thank you very much for listening to my webinar and if you have any questions or comments, please let me know and look forward to discussing it with you further.