Battlefield Pharma
Tens of thousands of soldiers are stationed on battlefields in remote locations where even the most common drugs can be hard to come by. Portable units that can synthesize both small- and large-molecule drugs on demand may sound like the work of science fiction, but efforts in this field are progressing at a surprising pace.
Drugs on Demand
Tyler McQuade has gone from chemistry professor to Deputy Director of DARPA’s Defence Sciences Office. Flow chemistry processes have been a continuous theme in his research. Here, he explains how clever chemistry can help make drugs on demand.
Drugs on Demand
Tyler McQuade has gone from chemistry professor to Deputy Director of DARPA’s Defence Sciences Office. Flow chemistry processes have been a continuous theme in his research. Here, he explains how clever chemistry can help make drugs on demand.
Making drugs on demand sounds like science fiction. At the outset, did you believe it would work?
I was actually one of the few people who thought it would be possible! Before I joined DARPA, I was the first recipient of funding under the PoD program, resulting in a modestly complex continuous synthesis system, using solid-supported reagents, which allowed end-to-end PoD-type synthesis of ibuprofen with decent purity and yield. To give you the history, DARPA’s medicines on demand effort was initiated by Geoff Ling (who served as served as the Director of DARPA’s Biological Technologies Office from 2014 until 2016). In one conversation I had with Geoff, he suggested developing a flexible synthetic system that could make every possible medicine from basic materials – such as pencil lead, eggshells, fertilizer and a sprinkle of metal! While it is true that those materials are sources of the key elements – carbon, sulphur, nitrogen and metals – I wanted to back up a little, and suggested starting with themes, such as focusing on limited types of reaction that would give a broad range of output. We soon demonstrated that you could take essentially the same reactions that were used for making ibuprofen and synthesize atropine, although we never published this.
Since then, our collaborators have brought a chemical engineer perspective to the project. For example, Klavs Jensen, Tim Jamison and Allan Myerson from Massachusetts Institute of Technology (MIT) pointed out that the number of unit operation types in drug synthesis is relatively small: heating, reagent addition over time, extractions, distillations, heterogeneous phase reactions, and so on. By mixing and matching these modular unit operations, you can achieve many outcomes – and this is the basis of the PoD system. At present, we swap the unit operations manually, but we’re creating an automated system that can reconfigure itself to run different chemistry on the fly, which is unprecedented.
Fluidics systems for continuous manufacturing are scalable in multiple ways: outwards, upwards and in terms of run times. There are lots of tricks that allow us to accommodate a much wider range of scales than people imagine. The current version is roughly the size of an under-counter refrigerator, but we can make the boxes smaller or larger. In Bio-MOD, we have made a handheld device that can produce a single dose, but we also have a bigger version that can make thousands of doses.
What is required to make these new technologies usable in the field?
The first hurdle is regulatory review. The FDA must be assured that drugs are produced in a verifiably safe way, and this could be challenging for distributed manufacturing systems. But I welcome that scrutiny – the agency’s rigorous standards have helped us visualize the future as GMP in a box, and work out how to create and monitor GMP standards in that environment. Manufacturing in a box actually has many advantages; for example, it is easier to control particle count than in a big factory. Also, we are borrowing concepts from biomanufacturing, such as disposable linings for reaction vessels to prevent cross-contamination, and removable parts to reduce impurities. In theory, you could make a reactor that is hermetically sealed from site of production to product implementation. We are addressing all the regulatory concerns right now. In fact, we’ve built a box specifically designed to be part of an FDA regulatory filing, and we’ll present data generated by this machine to the FDA in 2018.
Next, we must enhance the PoD system’s capabilities so that it can make more complex molecules. At present, molecules with challenging chemistry, such as atropisomers, structures with 10 stereocenters, or really congested quaternary centres, are still beyond us. And some reactions that are trivial in batch processes remain problematic in our system. For example, for convergent syntheses, we must develop processes with two parallel trains, so that intermediate A is synthesized in one train and intermediate B in the other, before combining the trains.
How do you envisage the future of drug manufacture?
In an ideal world, when a patient visits the doctor, his genome would be quickly sequenced, and perhaps also screened at the epigenetic level. The information would be sent to the drug synthesizer in the doctor’s office, which would immediately make the perfect drug for the patient. From my point of view, the exciting aspect of all the work in this area is that it could significantly improve the quality of medicine, while at the same time opening up new ways of interacting with patients and improving safety for the people who actually make medicines.
There may be bumps in the road, of course. But the people in the pharma industry are among the smartest I know, and I am certain that they will be able to adjust to this new reality and embrace it. Another difficulty may be that the market is just not ready for these developments. In fact, I often liken these technologies to the first television. When the cathode ray tube was first assembled into a machine to disseminate pictures, it was in an uncomfortable marketing position: why would anybody want a television when there was no content for it, and why would you create television content if nobody had one? We are in a similar position now with medicine on demand. Of course, it’s hard to envisage this kind of system because it’s so new, and people are sceptical because they hear so much hype about the future (personally, I am still waiting for somebody to make a flying car). But our work is gaining traction and even the FDA believes it will be an important part of medicine manufacture.
What are the challenges of delivering drugs to the battlefield?
On the battlefield, doctors do not have access to all the resources and medicines they would in a normal hospital – and if you run out of a medicine you can’t just request more stock and expect it to arrive quickly. It’s very frustrating for physicians, but it’s simply not possible to get everything they might require to the frontlines. Even with drop-shipping and helicopters, it can’t be done; cargo space is limited.
Also, battlefield logistics is associated with a lot of wasted medicines. For example, chemical warfare antidotes must be carried at all times because if troops are exposed they must be treated immediately. But once the medicines are out of date, they are discarded. Ultimately, this means that a large quantity of military-specific drugs are being bought, transported and stored in case of a very low-probability event, and then thrown away. It would be better to have just a small amount of drugs on standby to kick-start the response to an emergency, and to have an on-demand machine to manufacture sufficient drug to cover any shortfall. This means that troops would be mainly stocking stable raw materials with an unlimited shelf-life, rather than an expensive drug with a relatively short shelf-life. It would eliminate a huge yearly cost.
What are DARPA’s main medicine-on-demand programs?
DARPA’s goal is to develop an on-demand APImanufacturing platform that can produce up to 20,000 doses per day. We have two major programs in this area: PoD and Bio-MOD. PoD is the most advanced project and has been running since 2010; Bio-MOD was created in 2012. It would be better to have a single box that could manufacture both biologics and small molecules, but the techniques are too dissimilar to make that work. Even for small molecules alone, compressing all the different fundamental unit operations into a single box has been challenging, but our collaborators have made significant progress in this field.
In 2015, we also introduced the “Make-It” initiative – the objective being to develop the ability to manufacture any compound from just a few precursors. Traditional small-molecule API manufacturing begins with raw materials that are then refined into intermediates, which, in turn, are subjected to transformations prior to being made into final products. For example, BP purifies raw materials and gives them to BASF, which refines them and gives them to Pfizer, which conducts transformations, and so on, until you reach the final product. Make-It asserts that this entire stream can fit into a box – an ambition which has been made possible by advances in synthetic organic chemistry and artificial intelligence (AI). The AI’s function is to apply organic chemistry knowledge and to design the optimum synthetic pathway from simple raw materials to any pharmaceutical product. Our partners have developed some amazing AI tools that are already equivalent to a well-trained post-doc in terms of the quality of the syntheses they design. We’re also developing hardware to carry out those syntheses.
Ultimately, we hope to develop a stand-alone system from which you can generate any molecule, whether new or known. The AI component will figure out how to make it, and the machine will produce it from a few simple raw materials.
Patients expect that the medicines they need will be available when they need them. In the industrialized world, with robust supply chains and advanced infrastructure, this expectation is usually met. In remote areas, however, it’s a different story – mainly due to the difficulties of transporting and appropriately storing medical supplies in the context of poor infrastructure. These issues are typically associated with extreme circumstances, such as natural disasters and epidemics, but they also apply to battlefields.
Accordingly, the Defense Advanced Research Projects Agency (DARPA) – the blue-sky research arm of the US military – has been investigating methods of overcoming the logistics barrier so that medicines can be reliably accessed by soldiers stationed in remote areas. DARPA’s vision is of portable devices that can rapidly synthesize FDA-approved drugs, as required, in any location. In pursuit of this goal, the agency is funding a number of researchers under its Pharmacy on Demand (PoD) and Biologically-derived Medicines on Demand (Bio-MOD) initiatives. PoD, which has already passed beyond the proof of principle stage, relies on miniaturization of known reactions in order to quickly and cost-effectively generate batches of small-molecule drugs from shelf-stable precursors. The focus of Bio-MOD, an equivalent system intended for the production of biologicals, is on the development of systems that can produce several therapeutic proteins from a single cell line, or cell-free system, in a device the size of a laptop.
Portable, on-demand capabilities would transform drug logistics in extreme environments, but the implications may also extend to the whole pharmaceutical manufacturing industry, enabling distributed manufacturing and making it economically feasible to manufacture a specific drug and dose according to the specific needs of each individual patient. Some even speculate that each pharmacy or doctor’s office may one day have its own API manufacturing capability.
How close are we to the real-world implementation of drugs-on-demand technology? To find out, we spoke to DARPA, as well as some of the researchers involved in this exciting field.
Drugs on Demand
Tyler McQuade has gone from chemistry professor to Deputy Director of DARPA’s Defence Sciences Office. Flow chemistry processes have been a continuous theme in his research. Here, he explains how clever chemistry can help make drugs on demand.
Tyler McQuade has gone from chemistry professor to Deputy Director of DARPA’s Defence Sciences Office. Flow chemistry processes have been a continuous theme in his research. Here, he explains how clever chemistry can help make drugs on demand.
How did you become interested in the idea of making medicines on demand?
Before joining DARPA in 2013, I spent many years in academia where I focused on synthetic organic chemistry, particularly catalysis technologies to enable new chemistries. As a result, I became very familiar with continuous processes in the context of flow chemistry. Like most academics, I expended a lot of time and effort to achieve tenure, but after reaching that point I decided that I was ready for something new, and I was delighted to have the opportunity to join DARPA. It’s a unique organization where they are happy for us to push the limits of creativity, providing that the work is groundbreaking with the potential to improve national security. DARPA reaches for transformational change instead of incremental advances. I started out as a program manager, before becoming Deputy Director of DARPA’s Defense Sciences Office in January 2017.
Before joining DARPA, I’d felt for some time that pharma manufacturers were ready for new manufacturing technologies, but that they needed somebody else to remove the major regulatory risks. DARPA’s interests in battlefield medicine seemed to go right to the heart of the problem – the need for more flexible manufacturing technologies. And DARPA is not the only organization working in this area – there are many other excellent research groups working in this field too – in particular, Steven Ley and Lee Cronin are outstanding participants; so I think our programs are part of a broader revolution in medicine making. Perhaps that is partly due to the similarities between the logistical challenges faced by both battlefield medicine and personalized medicine. Personalized drugs specific to a given patient may be theoretically feasible, but unavailable in practice because of logistical or cost constraints – it’s little different from a battlefield scenario.
What are the challenges of delivering drugs to the battlefield?
On the battlefield, doctors do not have access to all the resources and medicines they would in a normal hospital – and if you run out of a medicine you can’t just request more stock and expect it to arrive quickly. It’s very frustrating for physicians, but it’s simply not possible to get everything they might require to the frontlines. Even with drop-shipping and helicopters, it can’t be done; cargo space is limited.
Also, battlefield logistics is associated with a lot of wasted medicines. For example, chemical warfare antidotes must be carried at all times because if troops are exposed they must be treated immediately. But once the medicines are out of date, they are discarded. Ultimately, this means that a large quantity of military-specific drugs are being bought, transported and stored in case of a very low-probability event, and then thrown away. It would be better to have just a small amount of drugs on standby to kick-start the response to an emergency, and to have an on-demand machine to manufacture sufficient drug to cover any shortfall. This means that troops would be mainly stocking stable raw materials with an unlimited shelf-life, rather than an expensive drug with a relatively short shelf-life. It would eliminate a huge yearly cost.
What are DARPA’s main medicine-on-demand programs?
DARPA’s goal is to develop an on-demand APImanufacturing platform that can produce up to 20,000 doses per day. We have two major programs in this area: PoD and Bio-MOD. PoD is the most advanced project and has been running since 2010; Bio-MOD was created in 2012. It would be better to have a single box that could manufacture both biologics and small molecules, but the techniques are too dissimilar to make that work. Even for small molecules alone, compressing all the different fundamental unit operations into a single box has been challenging, but our collaborators have made significant progress in this field.
In 2015, we also introduced the “Make-It” initiative – the objective being to develop the ability to manufacture any compound from just a few precursors. Traditional small-molecule API manufacturing begins with raw materials that are then refined into intermediates, which, in turn, are subjected to transformations prior to being made into final products. For example, BP purifies raw materials and gives them to BASF, which refines them and gives them to Pfizer, which conducts transformations, and so on, until you reach the final product. Make-It asserts that this entire stream can fit into a box – an ambition which has been made possible by advances in synthetic organic chemistry and artificial intelligence (AI). The AI’s function is to apply organic chemistry knowledge and to design the optimum synthetic pathway from simple raw materials to any pharmaceutical product. Our partners have developed some amazing AI tools that are already equivalent to a well-trained post-doc in terms of the quality of the syntheses they design. We’re also developing hardware to carry out those syntheses.
Ultimately, we hope to develop a stand-alone system from which you can generate any molecule, whether new or known. The AI component will figure out how to make it, and the machine will produce it from a few simple raw materials.
Making drugs on demand sounds like science fiction. At the outset, did you believe it would work?
I was actually one of the few people who thought it would be possible! Before I joined DARPA, I was the first recipient of funding under the PoD program, resulting in a modestly complex continuous synthesis system, using solid-supported reagents, which allowed end-to-end PoD-type synthesis of ibuprofen with decent purity and yield. To give you the history, DARPA’s medicines on demand effort was initiated by Geoff Ling (who served as served as the Director of DARPA’s Biological Technologies Office from 2014 until 2016). In one conversation I had with Geoff, he suggested developing a flexible synthetic system that could make every possible medicine from basic materials – such as pencil lead, eggshells, fertilizer and a sprinkle of metal! While it is true that those materials are sources of the key elements – carbon, sulphur, nitrogen and metals – I wanted to back up a little, and suggested starting with themes, such as focusing on limited types of reaction that would give a broad range of output. We soon demonstrated that you could take essentially the same reactions that were used for making ibuprofen and synthesize atropine, although we never published this.
Since then, our collaborators have brought a chemical engineer perspective to the project. For example, Klavs Jensen, Tim Jamison and Allan Myerson from Massachusetts Institute of Technology (MIT) pointed out that the number of unit operation types in drug synthesis is relatively small: heating, reagent addition over time, extractions, distillations, heterogeneous phase reactions, and so on. By mixing and matching these modular unit operations, you can achieve many outcomes – and this is the basis of the PoD system. At present, we swap the unit operations manually, but we’re creating an automated system that can reconfigure itself to run different chemistry on the fly, which is unprecedented.
Fluidics systems for continuous manufacturing are scalable in multiple ways: outwards, upwards and in terms of run times. There are lots of tricks that allow us to accommodate a much wider range of scales than people imagine. The current version is roughly the size of an under-counter refrigerator, but we can make the boxes smaller or larger. In Bio-MOD, we have made a handheld device that can produce a single dose, but we also have a bigger version that can make thousands of doses.
What is required to make these new technologies usable in the field?
The first hurdle is regulatory review. The FDA must be assured that drugs are produced in a verifiably safe way, and this could be challenging for distributed manufacturing systems. But I welcome that scrutiny – the agency’s rigorous standards have helped us visualize the future as GMP in a box, and work out how to create and monitor GMP standards in that environment. Manufacturing in a box actually has many advantages; for example, it is easier to control particle count than in a big factory. Also, we are borrowing concepts from biomanufacturing, such as disposable linings for reaction vessels to prevent cross-contamination, and removable parts to reduce impurities. In theory, you could make a reactor that is hermetically sealed from site of production to product implementation. We are addressing all the regulatory concerns right now. In fact, we’ve built a box specifically designed to be part of an FDA regulatory filing, and we’ll present data generated by this machine to the FDA in 2018.
Next, we must enhance the PoD system’s capabilities so that it can make more complex molecules. At present, molecules with challenging chemistry, such as atropisomers, structures with 10 stereocenters, or really congested quaternary centres, are still beyond us. And some reactions that are trivial in batch processes remain problematic in our system. For example, for convergent syntheses, we must develop processes with two parallel trains, so that intermediate A is synthesized in one train and intermediate B in the other, before combining the trains.
How do you envisage the future of drug manufacture?
In an ideal world, when a patient visits the doctor, his genome would be quickly sequenced, and perhaps also screened at the epigenetic level. The information would be sent to the drug synthesizer in the doctor’s office, which would immediately make the perfect drug for the patient. From my point of view, the exciting aspect of all the work in this area is that it could significantly improve the quality of medicine, while at the same time opening up new ways of interacting with patients and improving safety for the people who actually make medicines.
There may be bumps in the road, of course. But the people in the pharma industry are among the smartest I know, and I am certain that they will be able to adjust to this new reality and embrace it. Another difficulty may be that the market is just not ready for these developments. In fact, I often liken these technologies to the first television. When the cathode ray tube was first assembled into a machine to disseminate pictures, it was in an uncomfortable marketing position: why would anybody want a television when there was no content for it, and why would you create television content if nobody had one? We are in a similar position now with medicine on demand. Of course, it’s hard to envisage this kind of system because it’s so new, and people are sceptical because they hear so much hype about the future (personally, I am still waiting for somebody to make a flying car). But our work is gaining traction and even the FDA believes it will be an important part of medicine manufacture.
<link issues protein-synthesis-but-not-as-we-know-it>Protein Synthesis – But Not as We Know It
