A comparability study asks whether material produced after a change behaves like material produced before it. That sounds straightforward until you notice how many things tend to change at once, and how few studies are designed so that only one of them did.
One variable, or you have measured nothing
The entire logic of a comparability exercise rests on everything except the change under test holding still. When a process change coincides with a new donor, a different format, a different operator or a different assay lot, the study reports a difference it cannot attribute.
This is the failure that recurs most, and it is usually not carelessness. It is a scheduling artefact: the change happens when it happens, and whatever material is available at that moment gets used.
What has to be held constant
| Variable | Why it confounds | How to hold it |
|---|---|---|
| Donor | Between-donor variation usually exceeds the effect under test | Same donors before and after, which requires recall or banked material |
| Material format | Fresh and cryopreserved behave differently | Same format both sides |
| Grade | Different inputs and quality systems | Change grade separately from process |
| Assay reagents | Lot-to-lot drift in antibodies and media | Reserve a single lot for the whole study |
| Operator and site | Technique differences are real and measurable | Same hands, or randomise deliberately |
| Timing after thaw | Functional state shifts across the rest period | Fixed interval, recorded |
Related product
Leukopak formats. Single-donor starting material, fresh or cryopreserved, with full donor documentation.
Decide the acceptance criteria first
Criteria written after results are available are not criteria, they are an interpretation. Before running anything, state which attributes will be compared, what magnitude of difference would be considered meaningful, and what the study will conclude in each outcome.
The difficult part is the second one, because a meaningful difference has to be defined against something. Historical assay variability across replicate runs on the same material is the usual reference point, and a program that has never characterized its own assay variability has no basis for deciding whether a difference matters.
Run it side by side where you can
Comparing new material against a historical dataset introduces every drift that occurred in between. Comparing them in the same run, on the same day, with the same reagents, removes most of that at the cost of needing pre-change material still in hand.
Which is an argument for retaining reference material deliberately rather than consuming everything as it arrives. A retained aliquot of the original lot, stored properly, is what makes side-by-side comparison possible a year later. Storage requirements are covered in vapor-phase storage.
Measure more than one thing
Identity, purity and viability establish that the material is what it claims to be. Function establishes that it does what it did. A comparability study resting on phenotype alone can pass while the material has changed functionally, which is the outcome that surfaces later and expensively.
Where the material feeds a process, running the process on both is stronger than characterising the input alone, because it tests the thing you actually care about.
Planning for it before you need it
Comparability is cheapest when anticipated. That means retaining reference material, characterising assay variability early, documenting donor identity alongside results rather than in procurement records, and establishing at first purchase whether the same donors will be obtainable later.
OrganaBio documents a repeat-collection program for eligible donors, scoped per program rather than guaranteed, with characterization applied at donor program level across leukopaks, PBMCs and isolated populations. The grade-change case is covered in moving from RUO to cGMP.
Frequently asked questions
What makes a comparability study valid?
That only the variable under test changed. When a process change coincides with a new donor, a different format, a different operator or a different assay lot, the study reports a difference it cannot attribute to anything.
Why is donor the most important variable to hold constant?
Because between-donor variation usually exceeds the effect being tested. If the same donors cannot be obtained on both sides of a change, the study measures process and donor together and no statistical treatment separates them afterwards.
When should acceptance criteria be set?
Before running anything. Criteria written after results are available are an interpretation rather than criteria. Define which attributes are compared, what magnitude of difference is meaningful, and what conclusion follows from each outcome.
How do I decide what difference is meaningful?
Against your own assay variability, measured across replicate runs on the same material. A program that has never characterized that variability has no basis for judging whether an observed difference matters.
Should comparability be run side by side or against historical data?
Side by side where possible, since comparing against a historical dataset imports every drift that occurred in between. That requires retaining pre-change reference material rather than consuming everything as it arrives.
Is phenotypic comparison sufficient?
No. Identity, purity and viability establish that material is what it claims to be; function establishes that it does what it did. A study resting on phenotype alone can pass while the material has changed functionally.
How do I plan for comparability before it is needed?
Retain reference material, characterize assay variability early, record donor identity alongside results rather than in procurement files, and establish at first purchase whether the same donors will be obtainable later.
Talk to OrganaBio
Working through this on a live program?
The scientific team works through sourcing and specification questions with cell therapy and research groups directly, including donor characterization, format selection and documentation scope.
