The worst case in a shared facility answers two separate questions: which product is hardest to remove from the equipment, and which product sets the tightest residue limit. The first is driven by solubility in the cleaning medium and by how the product behaves on surfaces; the second by its health-based exposure limit (PDE) relative to the next product's batch size and dose. Often they are different products. A defensible matrix approach validates the cleaning procedure on the hardest-to-clean product and applies the tightest limit across the group, or covers both with two markers, and documents why every other product is covered.
Why does one axis not work?
Choosing only by hazard picks the most potent product even if it dissolves instantly in the cleaning agent; the validation then proves very little about the products that stick. Choosing only by solubility picks a product that is hard to remove but may have a generous limit; the validation then passes against a limit that is meaningless for the potent product. The EMA guideline on health-based exposure limits for shared facilities (2014) and the subsequent Q&A make the hazard side mandatory; Annex 15 and the industry guides from APIC and ISPE (Risk-MaPP) make the cleanability side explicit. The worst case sits where both meet.
Step 1: collect the data per product
For each product on the shared equipment train you need:
- PDE (or another health-based exposure limit) in mg/day, set and signed by a qualified toxicologist
- Solubility in the cleaning medium at the cleaning temperature, measured or, for triage, estimated
- Maximum daily dose and minimum batch size of every product that could follow
- Practical cleanability information: residue behaviour, formulation (tackiness, film formers, coatings), lab coupon tests if available
Step 2: compute the limit each product would set
The maximum allowable carryover into the next product B:
MACO = PDE(A) × MBS(B) / MDD(B)
with MBS the minimum batch size of B and MDD its maximum daily dose. Dividing by the shared product-contact surface gives the surface limit; multiplied by the swab area it gives the amount per swab.
A worked example with four products
Shared train with 180 m² of product-contact surface, next product with a minimum batch size of 150 kg and a maximum daily dose of 800 mg, swab area 25 cm². PDE values come from the toxicologist; solubilities are in water at 60 °C. The numbers are illustrative.
| Product | PDE (mg/day) | Solubility (mg/mL) | MACO (g) | Limit (µg/cm²) | Per swab (µg) |
|---|---|---|---|---|---|
| P1 | 0.010 | 8.0 | 1.9 | 1.04 | 26 |
| P2 | 2.5 | 0.02 | 469 | 260 | 6,510 |
| P3 | 0.30 | 0.4 | 56 | 31 | 781 |
| P4 | 0.08 | 0.9 | 15 | 8.3 | 208 |
Two answers come out of this table:
- Tightest limit: P1. With a PDE of 0.010 mg/day it allows 1.04 µg/cm², the lowest by a factor of eight.
- Hardest to remove: P2. At 0.02 mg/mL it is 400 times less soluble than P1. Its own calculated limit of 260 µg/cm² is far above the level at which residues become visible, so for P2 the visually clean criterion governs, not the MACO.
Step 3: choose and justify the marker
Three defensible options exist for this train:
1. Validate with P2 as marker against the P1 limit. If the cleaning procedure removes the least soluble product to 1.04 µg/cm², it will remove the others at least as well, provided they do not have worse surface behaviour. This is the classic matrix argument and needs the analytical method for P2 to reach 26 µg per swab with recovery data. 2. Two markers. Validate with P2 for cleanability and show P1 below its own limit in a separate run. More work, but robust if P1's formulation is unusual. 3. Group and bracket. If the train carries many products, group them by cleaning procedure and formulation type and pick a worst case per group.
Whatever you choose, the justification belongs in the cleaning validation master plan: the data per product, the ranking, the reason the marker covers the rest, and the trigger for re-evaluation when a new product joins the train.
Where does the time go, and what can be prepared?
The slow parts are almost always the same: PDE monographs for products without one, and solubility data for products where the cleaning temperature or medium differs from the literature. A first estimate of solubility and of the hazard profile from structure is enough to rank candidates and to decide which PDE monographs and which lab solubility measurements are needed first. It does not replace either. The PDE stays with the toxicologist, and a predicted solubility is a triage value, not a validation input.
Limits of the statement
- The PDE is the toxicologist's decision. Ranking tools and estimates can prioritise the work, never set the value.
- Solubility in water is a proxy. Detergents, pH and temperature of the actual cleaning step can change the ranking; for close calls, measure in the real medium.
- Formulation matters. A product with moderate solubility in a film-coated, tacky matrix can be harder to remove than the pure API suggests. Coupon tests settle this.
- Genotoxic and highly sensitising products follow additional rules and may need dedicated equipment regardless of the calculation.
Frequently asked questions
How is the worst-case product for cleaning validation selected?
By combining hazard and cleanability. Compute the residue limit each product would set from its PDE, rank cleanability by solubility in the cleaning medium and practical behaviour, and choose a marker that covers both: typically the hardest-to-clean product validated against the tightest limit of the group.
How do you calculate MACO from the PDE?
MACO = PDE of the previous product × minimum batch size of the next product / maximum daily dose of the next product. In the example, 0.010 mg/day × 150 kg / 800 mg/day gives 1.9 g, or 1.04 µg/cm² over 180 m².
Is the most toxic product always the worst case?
No. It sets the tightest limit, but if it is easily soluble in the cleaning medium, a validation run with it says little about products that are harder to remove. The marker has to address both questions.
What happens if the calculated limit is above the visible residue level?
Then the visually clean criterion governs. The equipment must be visually clean, and the calculated limit becomes irrelevant for that product.
Can predicted solubility be used in cleaning validation?
For ranking and planning, yes: it shows where measurements are needed first. For the validation itself, the solubility in the actual cleaning medium should be measured or supported by data.