The Future of ATMPs

July 2025

In part one of our two-part series on the future of ATMPs, David Phasey, Business Development Director at 3P Innovation, joins the podcast to discuss the drivers of the high cost-of-goods (COGs) for ATMPs and how automation can help reduce these COGs in the future.  

• Guests

  

  David Phasey
  

  Business Development Director

  3P Innovation

• Transcript

  Back to Top

  Download Transcript

  ---

  1 
  00:00:00,000 --> 00:00:04,000 
  Welcome to the ISPE podcast, Shaping the Future of Pharma,

  2
  00:00:04,000 --> 00:00:09,000 
  where ISPE supports you on your journey, fueling innovation, sharing insights,

  3 
  00:00:09,000 --> 00:00:14,000 
  thought leadership, and empowering a global community to reimagine what's possible.

  4 
  00:00:14,000 --> 00:00:21,000
  Hello and welcome to the ISPE podcast, Shaping the Future of Pharma. I'm Bob Chew, your host,

  5 
  00:00:22,000 --> 00:00:27,000 
  and today we have another episode where we'll be sharing the latest insights and thought

  6 
  00:00:27,000 --> 00:00:34,000 
  leadership on manufacturing technology, supply chains, and regulatory trends impacting the

  7 
  00:00:34,000 --> 00:00:40,000 
  pharmaceutical industry. You will hear directly from the innovators, experts, and professionals

  8 
  00:00:40,000 --> 00:00:47,000 
  driving progress and shaping the future. Thank you again for joining us, and now

  9 
  00:00:47,000 --> 00:00:56,000 
  let's dive into this episode. Our topic today is the future of advanced therapeutic medicinal

  10 
  00:00:56,000 --> 00:01:04,000 
  products, or ATMPs. To share more about this topic, I would like to welcome David Phasey,

  11 
  00:01:04,000 --> 00:01:12,000 
  Business Development Director at 3P Innovation, who recently presented at the 2025 ISPE

  12 
  00:01:12,000 --> 00:01:18,000 
  Europe Annual Conference in London. David, welcome to the podcast. We're glad to have you with us.

  13 
  00:01:19,000 --> 00:01:24,000 
  Thank you, Bob. It's great to be here, and hopefully a good topic to discuss.

  14 
  00:01:25,000 --> 00:01:33,000 
  Great, and I recall your presentation was well-received with a lot of good questions

  15 
  00:01:33,000 --> 00:01:38,000 
  from the audience, so we'll follow up with a few additional questions here.

  16 
  00:01:38,000 --> 00:01:46,000 
  So, David, in your opinion, what are the biggest drivers of the high cost of these advanced

  17 
  00:01:46,000 --> 00:01:53,000 
  therapies? Is it low volume? Is it a highly manual process? Is it the discovery costs?

  18 
  00:01:54,000 --> 00:02:01,000 
  Is it the supply chain logistics? Is it the therapeutic value to the patient, or something

  19 
  00:02:01,000 --> 00:02:09,000 
  else? Yeah, this is a really good question in that ATMPs are, of course, a broad range of different

  20 
  00:02:09,000 --> 00:02:18,000 
  therapy types, very different manufacturing chains, and as I've been researching this topic

  21 
  00:02:18,000 --> 00:02:22,000 
  conducting literature reviews and seeing what available information there is,

  22 
  00:02:23,000 --> 00:02:28,000 
  unsurprisingly, there isn't necessarily a completely fixed answer to the question.

  23 
  00:02:30,000 --> 00:02:36,000 
  So, I work, of course, in the automation side of the industry rather than a therapy developer

  24 
  00:02:36,000 --> 00:02:43,000 
  myself, and I'm aware that a number of large pharma companies, BMS, for example, have done

  25 
  00:02:43,000 --> 00:02:48,000 
  value stream mapping themselves to look at the manufacturing costs of their own products,

  26 
  00:02:50,000 --> 00:02:56,000 
  and, of course, they will have a cost model specific to their product. So, the way in which

  27 
  00:02:57,000 --> 00:03:03,000 
  I and the ISP team chose to look at this topic was perhaps rather than necessarily the biggest

  28 
  00:03:03,000 --> 00:03:10,000 
  drivers of cost, some of those reasonably well sort of understood and documented already in terms

  29 
  00:03:10,000 --> 00:03:17,000 
  of the drug discovery and the increase in cost of those. There's other good studies out there

  30 
  00:03:17,000 --> 00:03:23,000 
  on transportation costs, for example. There's a couple of items that you highlighted in your

  31 
  00:03:23,000 --> 00:03:30,000 
  question. So, the way in which we sort of try to rephrase that for ourselves was actually where

  32 
  00:03:30,000 --> 00:03:35,000 
  do we feel that as a community we can add the most value? So, where is actually the greatest

  33 
  00:03:35,000 --> 00:03:45,000 
  opportunity to reduce cost? And it struck us that there's already a reasonable degree of

  34 
  00:03:46,000 --> 00:03:51,000 
  effort being put into understanding transportation costs, and, of course, a lot of that came out of

  35 
  00:03:52,000 --> 00:03:59,000 
  recent events. So, instead, actually, what we focused on and where we saw the greatest

  36 
  00:03:59,000 --> 00:04:06,000 
  opportunity for reducing cost is in and around the manufacturing processes itself. So, you

  37 
  00:04:06,000 --> 00:04:15,000 
  mentioned in there manual processes, and really there is a significant cost associated with the

  38 
  00:04:15,000 --> 00:04:22,000 
  largely manual and semi-automated processes which have historically been laboratory-based,

  39 
  00:04:22,000 --> 00:04:29,000 
  and that transition to GMP does a significant cost with that. And so, that's where we decided

  40 
  00:04:29,000 --> 00:04:37,000 
  really to focus in on and explore because that's what we saw as an engineering community where we

  41 
  00:04:37,000 --> 00:04:43,000 
  felt there was most opportunity to add value and contribute to that reduction in cost.

  42 
  00:04:43,000 --> 00:04:51,000 
  Okay. So, my limited understanding, you've got a highly manual process, so you have

  43 
  00:04:53,000 --> 00:05:02,000 
  training of the people, you have the cost of the manpower, but I'm also aware that there's a pretty

  44 
  00:05:02,000 --> 00:05:09,000 
  significant effort around a myriad number of deviations that typically occur from a manual

  45 
  00:05:09,000 --> 00:05:19,000 
  process. Are there other aspects, and what's the low-hanging fruit here in the manual process

  46 
  00:05:20,000 --> 00:05:25,000 
  to implement automation and have the biggest positive cost reduction impact?

  47 
  00:05:26,000 --> 00:05:33,000 
  Yes. So, the way that we've been looking at that is the answer is perhaps a little different

  48 
  00:05:33,000 --> 00:05:39,000 
  depending on which the scale of manufacturing and the type of the ATMPs, but if we consider

  49 
  00:05:41,000 --> 00:05:48,000 
  probably the most common examples in describing that clean room, that manufacturing suite,

  50 
  00:05:48,000 --> 00:05:55,000 
  as you talked about, they're largely manual processes with individual pieces of automation

  51 
  00:05:55,000 --> 00:06:04,000 
  for individual process steps. Now, in that type of environment, surprisingly, it may actually be

  52 
  00:06:04,000 --> 00:06:09,000 
  that the automation is not the mechanization of process steps, and I think that's where a lot of

  53 
  00:06:09,000 --> 00:06:16,000 
  people would traditionally think of automation is actually the mechanization and that being the

  54 
  00:06:16,000 --> 00:06:26,000 
  opportunity. We see a lot about robotics, and that's very exciting, but when the process is,

  55 
  00:06:26,000 --> 00:06:34,000 
  you actually have a therapy SAT incubating or whatever the process step is within a piece of

  56 
  00:06:34,000 --> 00:06:43,000 
  equipment for a significant period of time, the mechanization doesn't have an immediate

  57 
  00:06:43,000 --> 00:06:49,000 
  transformational impact, and therefore, actually, when we're looking at those lower scales, the

  58 
  00:06:49,000 --> 00:06:58,000 
  opportunities appear to be more around process optimization and increase in quality, and really,

  59 
  00:06:58,000 --> 00:07:04,000 
  I think that comes about through instrumentation and therefore digitization of the processes,

  60 
  00:07:04,000 --> 00:07:08,000 
  and some of the tools to support that, the automation might be as simple as, for example,

  61 
  00:07:08,000 --> 00:07:18,000 
  a barcode scanner, the ability to streamline processes. I've done some work in a completely

  62 
  00:07:18,000 --> 00:07:25,000 
  parallel, but interesting sector of sterile instrument processing, and there are a number of

  63 
  00:07:25,000 --> 00:07:31,000 
  interesting observations in that, in that from the outside, when you first look at it, you would say

  64 
  00:07:31,000 --> 00:07:36,000 
  this can't be automated, it is really difficult, there are lots of unique process steps, it's

  65 
  00:07:36,000 --> 00:07:44,000 
  largely manual, you can't really automate this, you're a bit stuck, and I've seen some really

  66 
  00:07:44,000 --> 00:07:52,000 
  superb examples of lean manufacturing thinking being brought into that sector, and having an

  67 
  00:07:52,000 --> 00:08:01,000 
  extraordinary impact on the output of a facility purely based upon lean manufacturing processes

  68 
  00:08:01,000 --> 00:08:08,000 
  of studying the utilization of people within that room, the utilization of tools,

  69 
  00:08:10,000 --> 00:08:17,000 
  tracking materials through that room, and thereby increasing utilization of the assets that already

  70 
  00:08:17,000 --> 00:08:24,000 
  exist, and having a closer eye on the quality of those processes of which instrumentation and

  71 
  00:08:24,000 --> 00:08:31,000 
  digitization can help. So, a different form of automation, I think, would be, could offer

  72 
  00:08:32,000 --> 00:08:36,000 
  people that ability to make an initial transformation change, and that's where I'd see that

  73 
  00:08:36,000 --> 00:08:44,000 
  that loafing hanging fruit being. So, I recall seeing presentations at conferences, probably in the last

  74 
  00:08:44,000 --> 00:08:55,000 
  year, two different ones, one was this sterile clean room in a box, it sat on a coffee table,

  75 
  00:08:55,000 --> 00:09:03,000 
  almost, it was a Swiss company, and it did some of the rudimentary manipulations

  76 
  00:09:04,000 --> 00:09:10,000 
  around cell and gene therapy, so that was automating the manual process, but the other

  77 
  00:09:10,000 --> 00:09:19,000 
  one that I recall was kind of like a time motion study, almost, it was done by some sort of

  78 
  00:09:20,000 --> 00:09:30,000 
  visual tool with AI, I'm not sure, but it studied the detailed hand motions of the operator,

  79 
  00:09:31,000 --> 00:09:41,000 
  picking up the tools, making the connections, and it allowed to optimize and simplify

  80 
  00:09:42,000 --> 00:09:46,000 
  what the operator was doing. Is that the kind of thing that you're talking about?

  81 
  00:09:46,000 --> 00:09:52,000 
  Absolutely, I mean, that can have an impact in both, from a quality perspective, and certainly

  82 
  00:09:52,000 --> 00:09:58,000 
  those type of tools are well understood, used for improving good aseptic process, and therefore

  83 
  00:09:59,000 --> 00:10:06,000 
  you can end up with a double impact on cost there, in that, yes, you can make an efficiency

  84 
  00:10:06,000 --> 00:10:13,000 
  improvement in the way in which the operators in the clean room are able to produce that product,

  85 
  00:10:14,000 --> 00:10:19,000 
  but probably the biggest gain is in yield, because if they're not making mistakes in the process,

  86 
  00:10:19,000 --> 00:10:24,000 
  and you're able to generate a higher yield with that batch, or not have any failures within the

  87 
  00:10:25,000 --> 00:10:32,000 
  batch of manufacturing, that obviously is far greater, or can be potentially far greater cost

  88 
  00:10:32,000 --> 00:10:39,000
  saving than efficiency, so you can end up with both, and yes, that's the type I'm referring to.

  89 
  00:10:39,000 --> 00:10:46,000 
  It perhaps doesn't sound quite as glamorous as implementing robotics and things, which you might

  90 
  00:10:46,000 --> 00:10:52,000 
  immediately visualize, but in answering your question around low-hanging fruit, it's those

  91 
  00:10:52,000 --> 00:11:00,000 
  small steps that can actually have maybe even tens of percent improvement in the efficiency

  92 
  00:11:00,000 --> 00:11:05,000 
  of a process, but without actually necessarily spending money in order to achieve it, because

  93 
  00:11:05,000 --> 00:11:10,000 
  that's one of the other factors I've seen frequently come into these arguments, is

  94 
  00:11:12,000 --> 00:11:19,000 
  the instinct is to actually spend money to save money, and you can fall into the trap of actually

  95 
  00:11:19,000 --> 00:11:24,000 
  having dis-economies of scale, where you actually introduce more automation, and the payback

  96 
  00:11:24,000 --> 00:11:30,000 
  doesn't really arrive, depending on the throughput of your manufacturing process.

  97 <
  00:11:30,000 --> 00:11:40,000 
  Now, I also know about AI being used for vial inspections and that kind of thing.

  98 
  00:11:40,000 --> 00:11:40,000 
  Yes.

  99 
  00:11:41,000 --> 00:11:50,000 
  What about a camera with AI behind it that is watching everything that the operator does,

  100 
  00:11:51,000 --> 00:12:06,000 
  and could it detect and then analyze deviations visually and resolve them

  101 
  00:12:06,000 --> 00:12:14,000 
  almost instantly as either something significant, hey, high probability you just introduced

  102 
  00:12:14,000 --> 00:12:23,000 
  contamination, or yeah, it was a deviation, but it really doesn't matter. Is that an application

  103 
  00:12:23,000 --> 00:12:25,000 
  of technology here?

  104 
  00:12:26,000 --> 00:12:35,000 
  Yeah, and it's interesting you ask that question, because I would have said only a few

  105 
  00:12:35,000 --> 00:12:41,000 
  weeks ago that that was an interesting future technology. What I hadn't realized until I went

  106 
  00:12:41,000 --> 00:12:47,000 
  to another presentation recently is that's now a current technology that exists. It's been tested.

  107 
  00:12:50,000 --> 00:12:57,000 
  I'm sure people can find that company, and I found it really interesting. It was doing exactly

  108 
  00:12:57,000 --> 00:13:05,000 
  as described there, is that by studying many, many sample images, and they were able to use

  109 
  00:13:05,000 --> 00:13:15,000 
  AI to say, well, if action X is acceptable, is good aseptic processing, and action Y is not,

  110 
  00:13:15,000 --> 00:13:19,000 
  and a good number of images of examples, the AI was then able to identify

  111 
  00:13:22,000 --> 00:13:25,000 
  actions which were not considered best practice.

  112 
  00:13:26,000 --> 00:13:37,000 
  Do we think that that technology is more, I'll say, bulletproof than a human

  113 
  00:13:37,000 --> 00:13:44,000 
  analyzing the same information and making determinations?

  114 
  00:13:45,000 --> 00:13:51,000 
  That's a good question. I think where I saw, when I saw this presentation, where I saw

  115 
  00:13:52,000 --> 00:14:03,000 
  immediate value is the fact that you can provide this ongoing monitoring of a process that isn't

  116 
  00:14:03,000 --> 00:14:10,000 
  perhaps practicable to have, in some instances, but in others not, where you have an operator

  117 
  00:14:10,000 --> 00:14:17,000 
  and then another operator watching over their shoulder. Therefore, it probably adds that

  118 
  00:14:17,000 --> 00:14:23,000 
  additional assurance, because a lot of these processes are particularly challenging to

  119 
  00:14:23,000 --> 00:14:26,000 
  perform and remember each step and perform everything.

  120 
  00:14:28,000 --> 00:14:36,000 
  Exactly right. What do you see as the current barriers to adoption of truly impactful,

  121 
  00:14:36,000 --> 00:14:41,000 
  from a cost perspective, automation into ATMP manufacturing?

  122 
  00:14:42,000 --> 00:14:49,000 
  So, in that last question, I was answering, sort of considering where we are today in a more

  123 
  00:14:50,000 --> 00:14:57,000 
  manual cleanroom and describing some of the advantages we can see from relatively small

  124 
  00:14:57,000 --> 00:15:01,000 
  changes. If we think now very much towards the future, and I think that's where you're going

  125 
  00:15:01,000 --> 00:15:08,000 
  with that question, perhaps a little, quite a way into the future, really the fundamental

  126 
  00:15:09,000 --> 00:15:12,000 
  change, as I see it, is actually driven by the consumables.

  127 
  00:15:13,000 --> 00:15:19,000 
  And this was something I was talking about in my presentation. If you look to other sectors,

  128 
  00:15:19,000 --> 00:15:27,000 
  whether in the pharmaceutical or elsewhere, part of the, or a key part of that journey,

  129 
  00:15:27,000 --> 00:15:37,000 
  has been around standardization of componentry. USB, for example, USB-C and the standardization

  130 
  00:15:38,000 --> 00:15:43,000 
  of that connector. Now, that's actually more from the, in that instance, from the consumer benefit

  131 
  00:15:43,000 --> 00:15:53,000 
  perspective. But in this instance, standardization of consumable components, I see as being one of

  132 
  00:15:53,000 --> 00:16:02,000 
  the key milestones in enabling cost reduction. There's reasons for that. Now, it's because,

  133 
  00:16:02,000 --> 00:16:14,000 
  one, you can actually allow a lot of mechanization style automation to be designed

  134 
  00:16:14,000 --> 00:16:22,000 
  around these common sets of components and consumables. That in itself promotes competition

  135 
  00:16:22,000 --> 00:16:27,000 
  within that space. But also, if you have consumables that are themselves designed to

  136 
  00:16:28,000 --> 00:16:33,000 
  be automated in their actuation, which historically they haven't been, they've

  137 
  00:16:33,000 --> 00:16:41,000 
  been designed in classic sterile connectors, are designed for dexterous use of thumbs, etc.,

  138 
  00:16:41,000 --> 00:16:48,000 
  not operated by a robot. So, I suspect we should hopefully see a transformational

  139 
  00:16:48,000 --> 00:16:54,000 
  change in consumables and that's unlocking mechanization style automation. And along

  140 
  00:16:54,000 --> 00:17:02,000 
  with that, as we get more standardization in consumables, and whether that be bioreactors

  141 
  00:17:02,000 --> 00:17:10,000 
  themselves, tubing sets, sterile connectors, filtration sets, etc., that should also enable

  142 
  00:17:11,000 --> 00:17:18,000 
  larger volumes of componentries and more assurance in supply chains that allows the cost of those

  143 
  00:17:18,000 --> 00:17:25,000 
  components to come down. So, you can end up with a snowboard effect as we start to coalesce

  144 
  00:17:25,000 --> 00:17:31,000 
  around more standardized componentry. At the moment, I see it that we're in a diversification

  145 
  00:17:31,000 --> 00:17:35,000 
  phase as everyone's exploring different technologies. And again, you see these

  146 
  00:17:35,000 --> 00:17:40,000 
  parallels happen in other industries. And there is good pattern of this in the industrialization

  147 
  00:17:40,000 --> 00:17:46,000 
  of other sectors. You go through this growth phase of diversification as people try and explore

  148 
  00:17:46,000 --> 00:17:50,000 
  what's the best way of solving this problem, what are all these best technologies.

  149 
  00:17:50,000 --> 00:17:57,000 
  But after a number of years, history shows us we always end up coalescing around a number of

  150 
  00:17:57,000 --> 00:18:05,000 
  key solutions which the industries unify around, coalesce around, and that then unlocks the next

  151 
  00:18:05,000 --> 00:18:09,000 
  phase. So, that's where I see it really. It's probably the consumable as the core.

  152 
  00:18:10,000 --> 00:18:17,000 
  Okay. In your talk regarding economies of scale, you mentioned autologous versus allergenic

  153 
  00:18:17,000 --> 00:18:25,000 
  therapies. And you also mentioned small versus large patient populations. Now, to a large degree,

  154 
  00:18:25,000 --> 00:18:32,000 
  aren't these factors beyond the control of the manufacturer and instead driven by the specific

  155 
  00:18:32,000 --> 00:18:39,000 
  disease indication and patient population and genetic profile in terms of batch size

  156 
  00:18:40,000 --> 00:18:48,000 
  and patient populations? Yes, absolutely right. And so, the reason for highlighting that was

  157 
  00:18:49,000 --> 00:18:59,000 
  really twofold. Firstly, it's my aim in the content I'm sort of producing here is I'm

  158 
  00:19:01,000 --> 00:19:08,000 
  wanting to share information with the community and seek additional input from the community.

  159 
  00:19:08,000 --> 00:19:15,000 
  Seek additional input from the community. And in this instance, I think there is a…

  160 
  00:19:18,000 --> 00:19:23,000 
  If I make the assertion that we are going to see a reduction in the cost of goods,

  161 
  00:19:23,000 --> 00:19:29,000 
  the natural question that comes immediately back is, well, to what extent? Where are we going to

  162 
  00:19:29,000 --> 00:19:34,000 
  see it? By how much? That's, of course, a very, very difficult question. So, there was a particular

  163 
  00:19:34,000 --> 00:19:43,000 
  table that I put up in the presentation where I'm trying to add a degree of clarity into what

  164 
  00:19:43,000 --> 00:19:51,000 
  does cost of goods actually mean? What are all of the drivers? And where does the manufacturer

  165 
  00:19:51,000 --> 00:19:59,000 
  of ATMPs compare and contrast with other sectors, but also other pharmaceutical products? And

  166 
  00:20:00,000 --> 00:20:07,000 
  the patient population is obviously a key component in that. And awareness of it is necessary in order

  167 
  00:20:07,000 --> 00:20:13,000 
  to understand what the limitations are. So, that's one aspect is really helping try and as we try and

  168 
  00:20:13,000 --> 00:20:20,000 
  pick apart some clarity. The other component is actually when I was talking about standardisation,

  169 
  00:20:22,000 --> 00:20:26,000 
  we're in this diversification stage. We're obviously diversification in terms of

  170 
  00:20:27,000 --> 00:20:33,000 
  number of therapy types, the manufacturing process associated with them, and the technologies

  171 
  00:20:33,000 --> 00:20:38,000 
  associated with it. So, there's diversification in many, many different areas. That means that

  172 
  00:20:38,000 --> 00:20:47,000 
  it is quite hard to standardise. So, by reflecting back this diversification, my aim is then to try

  173 
  00:20:47,000 --> 00:20:54,000 
  and help the community identify where we can start seeing these pockets of similarity. Because

  174 
  00:20:54,000 --> 00:21:02,000 
  the sooner we start identifying where similarities exist, in therapy types A, B, and C,

  175 
  00:21:02,000 --> 00:21:07,000 
  whilst they are wildly different, actually, there's commonalities in these areas. If we can unify

  176 
  00:21:07,000 --> 00:21:13,000 
  manufacturing processes in there, then we can then effectively increase the size of the patient

  177 
  00:21:13,000 --> 00:21:20,000 
  population. Not by changing the therapy itself, which of course, as manufacturers, we can't do,

  178 
  00:21:21,000 --> 00:21:27,000 
  but by creating commonality within pockets of the manufacturing process. I realise that's a little

  179 
  00:21:28,000 --> 00:21:32,000 
  abstract in my description, but hopefully that makes sense.

  180 
  00:21:32,000 --> 00:21:43,000 
  So, what about distributed manufacturing, or basically the idea of a clean room trailer

  181 
  00:21:43,000 --> 00:21:51,000 
  in the hospital parking lot that has the ability to produce any number of these

  182 
  00:21:52,000 --> 00:21:58,000 
  ATMPs on demand? Is that feasible in the future, do you think?

  183 
  00:21:59,000 --> 00:22:03,000 
  Yes, feasible, of course, opens up a number of different ways of looking at that from a

  184 
  00:22:03,000 --> 00:22:11,000 
  technological perspective. I don't see any technological reason why that is not possible.

  185 
  00:22:11,000 --> 00:22:15,000 
  We've already seen some examples of people doing that kind of approach.

  186
  00:22:17,000 --> 00:22:23,000 
  The challenges probably come more, actually, when considering costs

  187 
  00:22:26,000 --> 00:22:31,000 
  about utilisation. It's something we haven't discussed in this podcast yet, but it is actually

  188 
  00:22:32,000 --> 00:22:38,000 
  one of the surprises to me, but probably not surprises to an accountant, as it were,

  189 
  00:22:38,000 --> 00:22:44,000 
  when looking at the costs of goods and where these costs come from. Utilisation is probably

  190 
  00:22:44,000 --> 00:22:54,000 
  one of the biggest cost drivers, in that if you have purchased all of this capital equipment in

  191 
  00:22:54,000 --> 00:23:01,000 
  the form of the trailer and the equipment that goes in it, the cost of the goods relies probably

  192 
  00:23:02,000 --> 00:23:08,000 
  on a high degree of utilisation of that equipment. Now, in conventional manufacturing,

  193 
  00:23:08,000 --> 00:23:16,000 
  if you're doing, say, a vial filling line, it's relatively easy to plan for the utilisation of

  194 
  00:23:16,000 --> 00:23:20,000 
  that. You can reasonably forecast how many batches you're going to produce a year, and therefore,

  195 
  00:23:20,000 --> 00:23:26,000 
  you can work out on a unit manufacturing basis what the cost of producing that individual vial is.

  196 
  00:23:26,000 --> 00:23:31,000 
  When we're talking about ATMPs, that becomes a considerably more difficult question,

  197 
  00:23:31,000 --> 00:23:36,000 
  and from some of the studies, as I've seen, that becomes one of the challenges, is that

  198 
  00:23:36,000 --> 00:23:47,000 
  you invest in this infrastructure, but you have far less control over the patient population demand

  199 
  00:23:48,000 --> 00:23:54,000 
  upon that service. And if you end up with a relatively low utilisation, you end up,

  200 
  00:23:55,000 --> 00:24:00,000 
  unfortunately, with a relatively high cost. And I think that's one of the areas

  201 
  00:24:02,000 --> 00:24:12,000 
  where I suspect, I very much hope, I'm sure we all hope, that the change in the style of the ATMPs,

  202 
  00:24:13,000 --> 00:24:20,000 
  the efficacy of them, actually means that we will end up moving more towards first-line treatments,

  203 
  00:24:20,000 --> 00:24:27,000 
  larger populations, and making these types of systems affordable in that you can achieve a

  204 
  00:24:27,000 --> 00:24:32,000 
  high utilisation from them because they become more norm. I suspect what we're going to see

  205 
  00:24:32,000 --> 00:24:38,000 
  in the industry is this snowballing type effect where one impacts the other.

  206 
  00:24:38,000 --> 00:24:46,000 
  Well, this has been a very interesting discussion. To recap, we talked about various ways that

  207 
  00:24:46,000 --> 00:24:57,000 
  automation could be introduced or expanded within the manufacture of ATMPs, not just robots,

  208 
  00:24:57,000 --> 00:25:03,000 
  which a lot of people think about, but how to supplement the human, make the human more

  209 
  00:25:04,000 --> 00:25:14,000 
  efficient, make the process more efficient, and all the quality system elements that go with it.

  210 
  00:25:14,000 --> 00:25:21,000 
  So, there's no single silver bullet, it seems. Would you agree with that?

  211 
  00:25:21,000 --> 00:25:26,000 
  I'm afraid so, yes. It is particularly complicated, but good opportunities at the same time.

  212 
  00:25:27,000 --> 00:25:32,000 
  Well, great. That brings us to the end of another episode of the ISPE podcast,

  213 
  00:25:33,000 --> 00:25:41,000 
  Shaping the Future of Pharma. A big thank you to our guest, David Fazey, for sharing his findings

  214 
  00:25:41,000 --> 00:25:46,000 
  on what the development of products in other sectors can tell us about how the potential

  215 
  00:25:46,000 --> 00:25:52,000 
  >reduction of cost of goods can be achieved in the future with respect to the manufacture of ATMPs.

  216 
  00:25:54,000 --> 00:25:59,000 
  Be sure to subscribe so you don't miss future conversations with the innovators,

  217 
  00:25:59,000 --> 00:26:05,000 
  experts, and changemakers driving our industry forward. On behalf of all of us at ISPE,

  218 
  00:26:06,000 --> 00:26:11,000 
  thank you for listening, and we'll see you next time as we continue to explore the ideas,

  219 
  00:26:11,000 --> 00:26:17,000 
  trends, and people shaping the future of pharma.