If you have a good process platform design, is quality by design necessary? It's an ongoing debate since the introduction of DOE studies and QbD as enforced by the FDA to ensure biologic product quality. There has been a significant surge in the intricacy of the design of process development based on these clear expectations of product quality set forth by the FDA.
A number of reasons can lead to the failure of a biologic drug candidate in the clinic, especially ones related to quality components. This has prompted the great need for early stage process validation, and the connectivity of process development parameters in order to assure a product's quality down the line. Changes can happen in storage, such as interactions with certain mediums that could affect your product and have an influence on the product's characteristics.
Considering that of all drug candidates that go through clinical development, roughly 5-10% of them will reach the end, quality by design is a compromise. Some argue you simply cannot cut corners rushing to get to proof of concept. The risk is too heavy, to spend the money and resources pushing past POC without complete knowledge of your product's characteristics, stability, safety, just to have your candidate fail in the clinic and not get approved.
It's important to take the time in the initial early stage development, so you can have greater assurance your product will be approved post proof of concept.
Furthermore, manufacturability of the product is a concern when early stage development has been too fast. You want to work out all problems that could arise as much as possible, before entering into manufacturing. If you move too quickly, without proper analytical knowledge of your product's stability, you're entering an almost "hail mary" situation once you get to manufacturing. Understanding scalability of your cells in a smaller setting allows for optimized large scale manufacturing.
It is undeniable that quality by design is a good concept, however overcoming the issues of cost, resources, and time when taking a QbD approach, is a challenge.
So the question is: How do you balance the front loaded approach with a back loaded one?