DOE and ICH Q8: Building the Design Space

Part of our complete guide to DOE for biologists.

ICH Q8 asks for something specific: not just a process that works, but documented understanding of why it works — which process parameters actually affect product quality, and across what range you can operate them while still meeting your quality attributes. That documented region is the Design Space: the multidimensional combination and interaction of inputs that's been demonstrated to provide quality assurance, not a single optimal recipe.

Why this can't come from tribal knowledge or OFAT

A process that "works" at one set of conditions doesn't tell you the boundaries of a Design Space — it tells you one point inside it, or possibly outside it if you got lucky. And testing parameters one at a time has the same blind spot in a regulatory context that it has everywhere else: if two Critical Process Parameters (CPPs) only affect a Critical Quality Attribute (CQA) in combination, an OFAT characterization study will miss that interaction entirely, and so will the Design Space boundary you draw from it. See DOE vs. OFAT for why that blind spot exists structurally, not just as a matter of running more OFAT trials.

How DOE actually builds the CPP → CQA map

This is normally a two-stage process, and it maps directly onto the same design choice covered in Which DOE Design Should I Use?:

Screen first. Start with every parameter that plausibly could be a CPP and run a screening design to find out which ones actually move your CQAs. This is where a lot of candidate parameters get ruled out cheaply, before you spend runs characterizing their exact effect.

Then characterize or optimize the ones that matter. For the parameters that survive screening, a response surface design maps the actual shape of the relationship between those CPPs and your CQAs across the range you intend to operate in — including interactions between CPPs and any curvature in the response. That mapped region, backed by the data, is what supports a Design Space claim.

What BioDOE does and doesn't do here

BioDOE generates the screening and response-surface designs behind this work and runs the statistical analysis — significance testing, interaction and curvature detection, and (on the Pro plan) full response-surface modeling for the optimization stage. That's real, useful groundwork for a Design Space submission.

What it doesn't do is replace the regulatory and quality expertise that turns that data into an actual submission. Deciding which parameters count as CPPs in the first place, how a Design Space claim gets documented and justified to a regulator, and how it's maintained post-approval are quality-system and regulatory-affairs judgment calls — BioDOE is a tool for the experimental and statistical work that feeds into them, not a substitute for that judgment, and it isn't certified or validated software for regulatory submission on its own.

Screening designs are available on the free Starter plan; the response-surface designs that map a full Design Space are a Pro feature.

Start free and screen your candidate CPPs — the response-surface step is there when you're ready for it.