As the title suggests, the FDA guidance is one year old and how is it doing? Has it created traction or is the industry still wondering what to do with it? Has it hit the terrible two's?
Well, it is not as active as a one year old going on two. It appears industry is still wrestling with what it all means. The three prong approach to the guidance is still there. And for those who might have forgotten, it includes the following:
A very strong process development program, read Quality by Design (QbD), is the precursor.
Demonstration of the process with qualification runs is still central stage. Translate qualification to mean validation.
After demonstration, move into continuous monitoring and read into that continuous improvement.
So where do we stand with these three elements? It is not unanimously agreed upon but at the recent Process and Analytical Validation Conference, sponsored by IBC in San Diego, February 27 and 28, there was vigorous discussion of the topics. Let's start with the least contentious topic.
3.) Everybody agreed that the idea to monitor processes and look for opportunity for improvement makes sense. Investment in process development, followed by validation activities, is a large amount of investment in resources so it made no sense to simply do the work and put the files on the shelf to gather dust. The activities should be value added and as such, let's get the most from them. As we monitor the progress of commercial operations, we can gather valuable information on process reliability, robustness and reproducibility. These data will lend themselves to analysis and recommendation for adjustments to the licensed process which could result in improved control strategies, improved inputs of raw materials and adjustments to the critical control points. All of these leads to improvement in the outputs, namely critical quality attributes (CQA). And then we are onto the continuous improvement life cycle. If all goes well then the processes will operate more reproducibly, and lead to improved yields, reliability and success. This should translate into lower cost of goods (COGS) and increased profitability. It also gives an opportunity to feedback to development a report of their success and how it might be handled differently in the future - read Lessons Learned. All of this sounds very optimistic. But all of this is dependent on another assumption. And that is that all this work will be possible. That is that when we discover an improved strategy for process control, an improved raw material or a new technological advance we will be able to incorporate these changes into the licensed process. In the past any change that is contemplated has been painful to implement even with comparability protocols and CBE 30's. And that brings us to the next point.
1.) To obtain all the advantages of point three, requires us to submit to the agency a full QbD package along with our submission. If we are a small molecular weight drug company, our odds are reasonably good. At present about 35 submissions have been made and at least 5 have been approved. Pfizer's Chantix and Selzentry has lead the way resulting in approval without a pre-approval inspection (PAI) for one molecule and a PAI with no observations related to the QbD submission for the other. In both cases the design space was accepted and internal change control processes were allowed to govern any process change within the design space. Ordinarily, any small change outside the routine operative area, would have required severe regulatory scrutiny including delays prior to implementation.
Biological companies have submitted full QbD packages but the number is far smaller than for the drug companies. To date about 6 have been made but no approvals have been made, at least to the author's knowledge. So where does that leave biotech companies? It leaves them behind the 8 ball. Let's assume that some are eventually approved. Well that puts them in the same place that drug manufacturers were about 3 year ago.
Perhaps the most telling comment was at the recent conference I referred to. Even process scientist that have made the submissions all agreed that it was painful, resource intensive and time consuming. They also indicated that the effort was Herculean. So why do they say that? If you examine a typical drug product, the number of unit operations that are involved is small: perhaps three or at most 5. Now if you examine a typical biotech product you are looking at 9-12 depending on the complexity. All of these steps must be considered as unit operations each with input, controls and outputs. This adds to the complexity of the PD work and hence to the QbD filing.
Those airing the opinion were from large biotech companies (Genentech and Pfizer). One panelist from a mid sized company (Shire) articulated that his company was taking a cautious approach to the path forward. Audience discussion included small companies articulating their concerns as well. In fact several articulated that QbD and hence the new PV guidance might be exceedingly difficult for small companies. Many of my clients articulate the same in private as they labor on what activities to do and what to delay or ignore.
Smaller companies, while not having the lavish pilot plants manned by platoons of process scientists with deep pockets do have something that large companies do not have. And that is lack of bureaucracy to allow rapid decision making and agility to the ever changing climate of development. They are part of a structure where risk is actually admired versus their counterparts that exhibit risk averse behaviour that is characterised by multilayered decision making. So I feel that the smaller companies that develop products might be able to compete with the leviathan biotechs because of their improved agility and risk appetite rather than by their resources alone or lack thereof. In fact they may never actually submit a full QbD package but may submit a partial, covering certain unit operations where flexibility has more value.
Smaller biotechs that intend to sell their product candidates or even themselves are immune, since they do note even enter this maelstrom of uncertainty. Many companies have approached me and have asked advice on how much, how soon and what.
Which leads us onto point 2.
2.) So what is to become of the process validation work that we have grown accustomed to, but I do not say grown to love. Will it go away? Or will it morph?
At present, after PD activity and scale up, we perform a ritual three runs (5 or 3 in the EU), pray for success and declare victory and file our NDA, BLA or MAA.
In the new paradigm, we have performed all the PD work at small scale which is a good model for large scale, so the three runs are really meaningless. We all agree that statistically, three runs do not add value. For statistical reproducibility, we require a larger number, say 20-30. But who is going to suggest to a regulator that instead of doing 3 runs and submitting, we do 20 and then submit? That will drive approval out another 2-3 months at minimum. No CEO, let alone CFO, will stand for it. And the regulators are not crazy enough to suggest that.
In fact, if you do a QbD submission correctly and have a validated scale down model and operate that small scale model within the constraints of large scale capability, there is no need for the 20-30 runs to demonstrate the robustness. Actually, the 3 runs would even become redundant. I would contend that all that is required is a commitment to review the first 20-30 runs at large scale to confirm that the small scale QbD work was valid but post approval. And if the scale down modeling was done right with appropriate confirmation, then the post approval confirmation is really icing on the cake. There should be no surprises at the commercial scale.
Now some of my ideas may sound outlandish or even revolutionary but I am able to express them. As a consultant, I do not have the regulatory sword of Damocles hanging over my head. I have no submission into the agency or planned for submission in the future like some of the readers. But because I am able to articulate the thoughts with no retribution, it might paint the future in a more friendly progressive manner.