Cleaning Validation — Risk-Based Approach That Survives Pre-Approval Inspection
Cleaning validation is one of the most frequently cited deficiency areas in FDA pre-approval inspections — not because companies have not validated their cleaning processes, but because the cleaning validation…
On this pageArticle overview
Cleaning Validation — Risk-Based Approach That Survives Pre-Approval Inspection
Cleaning validation is one of the most frequently cited deficiency areas in FDA pre-approval inspections — not because companies have not validated their cleaning processes, but because the cleaning validation strategy does not withstand the specific scrutiny FDA applies when a new product is being added to a multi-product facility or when an existing validation predates current regulatory expectations.
That distinction matters enormously. When an investigator walks into your facility during a pre-approval inspection and pulls your cleaning validation package, they are not asking whether you have a cleaning procedure. They are asking whether your acceptance criteria are scientifically defensible, whether your analytical method can actually detect residues at those acceptance criteria, and whether your program accounts for the potency and toxicology of every product that has ever been manufactured on that shared equipment. In my 25 years working GMP manufacturing and regulatory submissions across FDA and EMA jurisdictions, the gap between “we have cleaning validation” and “our cleaning validation survives pre-approval scrutiny” is where most programs fall short — and almost always on the same three dimensions.
The Shift to Health-Based Exposure Limits — Why the 10 ppm Rule Is No Longer Sufficient
The regulatory landscape for cleaning validation acceptance criteria shifted decisively with the publication of the EMA Guideline on Setting Health-Based Exposure Limits (EMA/CHMP/CVMP/SWP/169430/2012, effective June 2015). Before that, the field operated largely under the framework laid out in FDA’s 1993 Guide to Inspections: Cleaning Validation, which proposed three widely used acceptance criteria: no more than 10 ppm of product A in product B, no more than 1/1000th of the minimum therapeutic dose of product A in the maximum daily dose of product B, and visually clean. These criteria were pragmatic for their era. They gave manufacturers a workable basis for acceptance limits when toxicological data were sparse and analytical capability was more limited than it is today.
The problem is that the 10 ppm rule and the 1/1000th therapeutic dose rule are not health-based. They are convention-based. For a highly potent active pharmaceutical ingredient — a cytotoxic agent, a hormone, a biologic with sub-microgram pharmacological activity — 10 ppm of carryover into a subsequent product may represent a meaningful patient exposure risk. Conversely, for a relatively benign excipient, a 10 ppm limit may be unnecessarily conservative, constraining manufacturing operations without delivering additional patient protection. Neither outcome reflects good risk management.
The EMA guideline resolved this by requiring that acceptance criteria for shared manufacturing equipment be derived from the Permitted Daily Exposure (PDE), a toxicologically derived threshold that represents the daily exposure to a residue that is unlikely to cause an adverse effect when experienced over a lifetime. The PDE is calculated from the No Observed Adverse Effect Level (NOAEL) — the highest tested dose at which no critical toxicological effect is observed — using a structured uncertainty factor approach drawn from Appendix 3 of ICH Q3C:
PDE = NOAEL × Weight Adjustment / (F1 × F2 × F3 × F4 × F5)
where F1 accounts for extrapolation between species, F2 for interindividual variability in humans, F3 for short-duration studies (repeat-dose toxicology studies of less than four weeks), F4 for severe toxicity such as non-genotoxic carcinogenicity, neurotoxicity, or teratogenicity, and F5 for cases where no NOAEL could be established and the calculation must instead rely on a lowest-observed-adverse-effect level (LOAEL), which introduces additional uncertainty that the factor is designed to offset. The resulting PDE is expressed in mg/day and is specific to each active substance and route of exposure. It is not a single number applied universally; it is a product-specific, toxicology-based threshold that must be established for every product manufactured on shared equipment.
From the PDE, the Maximum Allowable Carryover (MACO) is calculated as:
MACO = (PDE × minimum batch size of next product) / maximum daily dose of next product
This formula ensures that even if the entire MACO limit is present in the subsequent batch and distributed uniformly, no patient receiving the maximum daily dose of the next product will exceed the toxicologically derived PDE of the preceding product. The minimum batch size and maximum daily dose are chosen deliberately as the conservative parameters — they represent the conditions under which carryover concentration in the finished product would be highest.
FDA has aligned with this direction. While the 1993 guide has not been formally withdrawn, FDA’s pre-approval inspection practice and its 2011 process validation guidance — which frames cleaning as subject to the same lifecycle, science-based rigor as manufacturing process validation — both reflect an expectation that health-based limits will be used where they can be established. When FDA investigators find a cleaning validation program that relies exclusively on the 10 ppm rule for potent compounds, without any PDE-based analysis, that finding will appear in the Form 483. I have seen it repeatedly, and the response required is not a simple administrative correction — it requires re-deriving MACO limits for every product-equipment combination, requalifying analytical methods against those limits, and potentially conducting additional cleaning studies.
The 10 ppm rule retains a role as a backup criterion where a PDE cannot be established — for cleaning agents, for degradation products without established toxicology, or for legacy situations where toxicological data are genuinely unavailable. But it is not a substitute for health-based limits in a program that will be reviewed during pre-approval inspection for a new product manufactured on shared equipment.
EMA’s own Q&A guidance on implementation (EMA/CHMP/CVMP/SWP/246844/2018, “Questions and answers on implementation of risk-based prevention of cross-contamination in production”) and ICH Q9(R1) on quality risk management reinforce this framework by explicitly requiring that the risk of cross-contamination in multi-product facilities be assessed using health-based limits and that the rationale for product grouping, worst-case product selection, and acceptance criteria derivation be documented with sufficient scientific rigor to support regulatory review. That 2018 Q&A document is explicit that HBELs are not automatically the cleaning acceptance limit — the acceptance criterion must still be justified through a documented risk assessment that accounts for cleaning process capability, not simply set at the calculated HBEL value.
21 CFR 211.67 — the underlying US cGMP regulation — requires that equipment and utensils be cleaned, maintained, and sanitized at appropriate intervals to prevent malfunctions or contamination that would alter the safety, identity, strength, quality, or purity of the drug product. The regulation does not specify an acceptance criterion. That specificity comes from regulatory guidance and from inspection expectations, which in 2025 mean PDE-based MACO derivation for any product with established toxicological data.
MACO Calculation and Analytical Verification — The Evidence Package FDA and EMA Expect
Understanding what MACO calculation and health-based limits require in practice means recognizing that the calculation is only the beginning of the evidence package. The number that comes out of the MACO formula is only as defensible as the analytical method used to verify that residues are below it — and this is where a large proportion of pre-approval inspection deficiencies actually originate.
The analytical method LOD must be at or below 50% of the calculated MACO. This is not arbitrary. If your MACO for compound A on a particular piece of equipment is 5 micrograms per swab, and your HPLC method has an LOD of 4 micrograms per swab, you cannot reliably confirm that residues are below your acceptance criterion. The method cannot see residues in the range that matters. FDA investigators will calculate this during inspection. They will take your MACO value, compare it to your method’s validated LOD, and if the LOD exceeds 50% of MACO, the analytical verification step in your cleaning validation is scientifically inadequate regardless of whether your swab samples returned values below the limit.
Swab recovery studies are a required component of the evidence package, and they must be conducted on the actual surface materials used in your equipment. Recovery factors for stainless steel differ from recovery factors for PTFE, polypropylene, or glass-lined vessels. The recovery correction factor must be applied to your swab results before comparing them to the MACO limit. If your swab recovery study showed 70% recovery and your uncorrected swab result was 2 micrograms, your corrected residue estimate is 2.9 micrograms — you cannot report the uncorrected number and compare it to your acceptance criterion. The ISPE Baseline Guide: Risk-MaPP provides detailed guidance on swab technique, recovery validation protocol, and the statistical treatment of recovery data that FDA expects to see.
Worst-case product selection requires documented scientific rationale. The worst-case product for cleaning difficulty is determined by physicochemical properties — a compound with high log P (lipophilicity) and low aqueous solubility is harder to clean from stainless steel surfaces than a hydrophilic compound. The worst-case product for patient risk is the compound with the lowest PDE — the most potent or most toxic substance on the equipment train. These two worst-case designations may not be the same product, and your cleaning validation protocol must address both dimensions explicitly. An equipment train that processes both a cytotoxic compound (low PDE, low acceptable carryover) and a waxy lipophilic excipient (high log P, difficult to clean) requires that cleaning studies be designed to demonstrate removal of the most difficult-to-clean compound at residue levels consistent with the MACO for the most potent compound. Failure to address this intersection is a frequent source of pre-approval inspection findings.
Visually clean remains a required check — 21 CFR 211.67 and both FDA and EMA guidances reference it explicitly — but it is a qualitative, supplementary observation, not an acceptance criterion. A surface that is visually clean may still carry residues well above the MACO limit for a potent compound. Documenting “visually clean” as the sole acceptance criterion in a cleaning validation protocol is not defensible in 2025 under either FDA or EMA pre-approval inspection expectations. Analytical verification at the MACO limit is required.
Equipment grouping through bracketing and matrixing strategies — validated under ICH Q9(R1) risk management principles — can reduce the total number of cleaning studies required without compromising the scientific rigor of the program. The rationale for grouping must be documented: which equipment share the same material of construction, geometry, surface finish, and cleaning procedure? Which compound within the group has the highest log P and the lowest PDE? The MACO for that compound, using that analytical method, validated for recovery on that surface material, is the acceptance criterion that covers the entire equipment group. Any new product added to the group requires re-evaluation of whether the worst-case designation is maintained.
Cleaning Validation Lifecycle Management — Revalidation Triggers and New Product Integration
Cleaning validation is explicitly classified as a lifecycle activity under both FDA’s process validation guidance and EMA’s shared facilities guideline. The phrase “ongoing verification” in the FDA 2011 process validation guidance applies to cleaning as directly as it applies to manufacturing process validation. A cleaning validation study conducted in 2015 does not remain valid indefinitely if the conditions under which it was conducted have changed.
The revalidation triggers that FDA and EMA expect to see in a cleaning validation lifecycle management protocol are specific and non-negotiable. Introduction of a new product to the equipment train requires re-evaluation of worst-case product selection and MACO calculations for all products already on the train, in addition to MACO calculation for the new product itself. Equipment modification — including replacement of a vessel, change of surface finish, change in configuration that affects cleanability — requires reassessment of whether the existing cleaning study data remain applicable to the modified equipment geometry. Change of cleaning agent requires new cleaning efficacy data and new analytical method validation for any cleaning agent residues subject to MACO limits. Manufacturing site change requires full revalidation under the new facility’s water quality, cleaning equipment, and procedure conditions. Batch size change that affects the contact surface area-to-batch-size ratio requires recalculation of MACO and reassessment of whether residue limits per unit surface area remain supportable.
The most operationally significant trigger in a growing CMO or multi-product development facility is the addition of new products. Every new IND compound that progresses to Phase 2 or Phase 3 manufacturing on shared equipment is a potential trigger for revalidation of the cleaning validation program for the entire equipment train — not just for the new compound itself. The new compound may have a lower PDE than any existing compound on the train. If it does, it becomes the new worst-case product for patient risk, and the entire MACO calculation framework must be recalibrated against it. This is not a hypothetical scenario. It is the exact situation that triggers Form 483 observations during pre-approval inspections when a sponsor submits an NDA or BLA for a potent compound and FDA finds that the facility’s cleaning validation was established against a less potent reference compound years earlier, with no documented reassessment following the introduction of the new compound.
The administrative infrastructure required to support lifecycle management is as important as the technical content of the validation studies themselves. Your cleaning validation master plan must define the responsible function for triggering re-evaluation, the documentation pathway for recording and approving the re-evaluation decision, and the timeline for completing any required additional validation work. When FDA requests your cleaning validation package during pre-approval inspection, they will review not only the validation studies but also the change control records, the re-evaluation decisions, and the evidence that the program has been actively maintained rather than left as a static historical file.
The integration of ICH Q9(R1) quality risk management principles into the cleaning validation lifecycle means that the decision about whether a particular change triggers full revalidation or documented re-evaluation without additional studies must be supported by a formal risk assessment. That risk assessment must be documented, scientifically grounded, and retained as part of the cleaning validation file. The risk assessment must address the specific change, the specific equipment, the specific products, and the specific acceptance criteria — generic risk management language that does not reference the actual MACO values and analytical method capabilities will not satisfy inspection expectations.
The XGene Health-Based Cleaning Validation Architecture
1. PDE-Based MACO Calculation for Every Product-Equipment Combination Derive PDE for each active substance manufactured on shared equipment using the EMA/CHMP/CVMP/SWP/169430/2012 framework. Calculate MACO using: MACO = (PDE × minimum batch size of next product) / maximum daily dose of next product. Document all input parameters with source references. Retain the 10 ppm rule only as a supplemental criterion where PDE cannot be established.
2. Worst-Case Product Selection with Documented Scientific Rationale Identify the worst-case product for cleaning difficulty (highest log P, lowest aqueous solubility) and the worst-case product for patient risk (lowest PDE) separately. Document the scientific basis for both designations. Confirm that cleaning studies are designed to verify removal of the most difficult-to-clean compound at acceptance criteria driven by the most potent compound.
3. Analytical Method Capability Verification Against MACO Limits Validate LOD for each analytical method and confirm LOD is at or below 50% of the calculated MACO for each product-equipment combination. Do not proceed to cleaning studies until analytical method capability is confirmed. Document the LOD-to-MACO ratio for each compound in the cleaning validation summary.
4. Swab Recovery Validation Across All Contact Surface Materials Conduct swab recovery studies on coupons representing each surface material in the equipment train (stainless steel grade, PTFE, borosilicate glass, polymers). Apply recovery correction factors to all swab sample results before comparison to acceptance criteria. Retain recovery study data as part of the permanent cleaning validation file.
5. Equipment Grouping Strategy for Bracketing and Matrixing Group equipment by material of construction, surface geometry, and cleaning procedure. Designate the worst-case equipment within each group. Validate that acceptance criteria for the worst-case equipment cover all members of the group. Document the scientific rationale for each grouping decision under ICH Q9(R1).
6. Lifecycle Management Protocol with Defined Revalidation Triggers Maintain a living cleaning validation master plan. Define revalidation triggers: new product introduction, equipment modification, cleaning agent change, manufacturing site change, batch size change affecting contact surface area. Assign ownership for monitoring and responding to triggers. Document all re-evaluation decisions and retain as part of the change control record.