Process validation: what Annex 15, the FDA and ICH Q7 actually say about the number of batches

Ask in an audit or inspection how many batches it takes to validate a manufacturing process and the answer is almost always the same: three. Open the four texts that actually govern the question and you find four different statements. EudraLex Volume 4, Annex 15 names three consecutive batches as generally acceptable and requires every number to be justified. ICH Q7 names three as a guide for active substances. The EMA guideline on process validation requires data on at least three production scale batches, unless otherwise justified, only where the dossier needs production scale data, for example for biologicals and non-standard processes; as a rule, pilot scale batches and a validation scheme for later execution at production scale suffice. And the FDA guidance Process Validation of January 2011 names no number at all. Keeping these four statements apart settles most discussions about the validation master plan, the PPQ protocol and the batch count faster than any rule of thumb.

Annex 15: the justification comes before the number

Annex 15 in its current version dates from March 2015 and has been in operation since October 1, 2015. Its legal basis is Art. 47 of Directive 2001/83/EC; it describes "the principles of qualification and validation which are applicable to the facilities, equipment, utilities and processes" and may serve as supplementary optional guidance for active substances without introducing additional requirements to Part II of the EU GMP Guide. The principle is a GMP requirement over the life cycle: "It is a GMP requirement that manufacturers control the critical aspects of their particular operations through qualification and validation over the life cycle of the product and process."

On the number of batches, the annex says two things that belong together. Section 5.19: "Each manufacturer must determine and justify the number of batches necessary to demonstrate a high level of assurance that the process is capable of consistently delivering quality product." Section 5.20: "Without prejudice to 5.19, it is generally considered acceptable that a minimum of three consecutive batches manufactured under routine conditions could constitute a validation of the process." An alternative number of batches may be justified, and the data from three batches may need to be supplemented with data from ongoing process verification. The order of the two sections is the actual content: the justification requirement comes first, and three is the default the text itself offers. A manufacturer running three batches is within what is expected; one running two or five needs a documented derivation. And even with three, the validation master plan should record why three is enough here.

The conditions for those batches sit right beside: validation batches "should be the same size as the intended commercial scale batches" (5.8), the equipment used is qualified and the analytical methods are validated (5.9). Which process parameters and quality attributes are critical must be justified and documented in advance, taking into account the results of any risk assessment (5.7). The protocol "defines the critical process parameters (CPP), critical quality attributes (CQA) and the associated acceptance criteria which should be based on development data or documented process knowledge" (5.21).

Four texts, three numbers and one blank

The blank in the FDA text is deliberate. The guidance defines process validation as "the collection and evaluation of data, from the process design stage through commercial production, which establishes scientific evidence that a process is capable of consistently delivering quality product" and recommends "objective measures (e.g., statistical metrics) wherever feasible and meaningful". The legal basis is Section 501(a)(2)(B) of the FD&C Act together with 21 CFR 211.100(a), which the guidance calls "the foundation for process validation", and 211.110(a), which requires manufacturers to validate the performance of those manufacturing processes that may be responsible for causing variability. The EMA guideline, in turn, governs only the data to be provided in the marketing authorization dossier; Annex 15 section 5.2.1 makes clear that the GMP requirements for process validation continue throughout the life cycle of the process. For a manufacturer with an EU authorization and FDA inspections, all three statements apply at once, and they do not contradict each other: three batches are the default in the EU, Annex 15 requires every number to be justified, and the FDA expects that justification to be statistical.

PQ is not the same PQ

A second point at which conversations between EU-trained and US-trained teams regularly talk past each other is the abbreviation PQ. In Annex 15 the qualification stages are called URS, DQ, FAT/SAT, IQ, OQ and PQ (3.2 to 3.14); IQ, OQ and PQ are therefore the end of the chain, not the whole of it. IQ includes verification of the installation against drawings and specifications, calibration and materials of construction (3.9). OQ includes "tests to confirm upper and lower operating limits, and/or 'worst case' conditions" (3.11); IQ and OQ may be combined as IOQ (3.10). PQ should normally follow the successful completion of IQ and OQ but may "in some cases" be performed in conjunction with OQ or process validation (3.13), with worst case batch sizes and a justified sampling frequency (3.14). The glossary defines PQ as "The documented verification that systems and equipment can perform effectively and reproducibly based on the approved process method and product specification". In Annex 15, PQ means the equipment.

In the FDA guidance, PQ is something else: the abbreviation for the whole of Stage 2, Process Qualification. "During the process qualification (PQ) stage of process validation", two elements are worked through: "(1) design of the facility and qualification of the equipment and utilities and (2) process performance qualification (PPQ)". PPQ is therefore the second element of Stage 2, and "A manufacturer must successfully complete PPQ before commencing commercial distribution of the drug product". A validation master plan that writes "PQ" without qualification means the equipment under Annex 15 and the process stage under the FDA model. The three stages of the guidance are Stage 1 Process Design, Stage 2 Process Qualification and Stage 3 Continued Process Verification.

Annex 15 itself defines process validation as "The documented evidence that the process, operated within established parameters, can perform effectively and reproducibly to produce a medicinal product meeting its predetermined specifications and quality attributes." That is the statement all stages lead up to.

Acceptance criteria are fixed before the first batch runs

The pitfall that both frameworks address explicitly concerns the order of criterion and result more than the number of batches. Annex 15 requires protocols to define "the critical systems, attributes and parameters and the associated acceptance criteria" (2.4). Any significant change to the approved protocol during execution, "e.g. acceptance criteria, operating parameters etc., should be documented as a deviation and be scientifically justified" (2.7). Results outside the predefined acceptance criteria are deviations and must be fully investigated (2.8); any subsequent change to acceptance criteria must be scientifically justified (2.9). Deriving acceptance criteria from the PQ data is therefore documented as an error in the text itself.

The FDA guidance says the same for the PPQ protocol. It states the manufacturing conditions, the data to be collected, the tests and acceptance criteria "for each significant processing step", the sampling plan, the statistical methods for intra-batch and inter-batch variability, and the handling of deviations; data should not be excluded from further consideration in terms of PPQ without a documented, science-based justification. The protocol is approved by the relevant departments and the quality unit before execution, and departures from the protocol "must be justified and approved … before implementation (§ 211.100)".

Validation does not end with the report

Both frameworks have a third phase, and both mean a program by it, not a document. Annex 15 knows three approaches to process validation: the traditional approach, continuous process verification as an alternative for products developed under quality by design principles (5.23, with a science based control strategy, PAT and multivariate statistical process control as tools, 5.24), and the hybrid approach, which requires "a substantial amount of product and process knowledge" (5.26). Ongoing process verification under 5.28 to 5.32 applies to all three: under an approved protocol, with a report, with "statistical tools … where appropriate", documented in the product quality review (5.32). The glossary equates the terms: "Ongoing Process Verification (also known as continued process verification)".

The FDA calls its Stage 3 Continued Process Verification and ties it to 21 CFR 211.180(e): "An ongoing program to collect and analyze product and process data that relate to product quality must be established". The data "should be statistically trended and reviewed by trained personnel"; the guidance recommends involving a statistician or a person trained in statistical process control in the data collection plan and the methods. Monitoring initially runs at the PPQ level, "until sufficient data are available to generate significant variability estimates", and is then "adjusted to a statistically appropriate and representative level". A CPV program kept as the closing document of the validation misses both texts.

Retrospective, concurrent, and the line between EU GMP and ICH Q7

Anyone validating "under EU GMP and ICH Q7" has to keep the two texts apart at one point. Annex 15 states twice, in the General section and in 5.3, that retrospective validation is "no longer considered an acceptable approach". ICH Q7, by contrast, knows three approaches in section 12.4: prospective as "the preferred approach" (12.41), concurrent for few or infrequent API batches or after a change to a validated process (12.43), and retrospective as an exception "for well established processes" under four conditions (12.44), with generally ten to thirty consecutive batches (12.50). For finished products under EU GMP, the retrospective route is closed; for established active substance processes, ICH Q7 keeps it open as an exception.

Concurrent validation, that is releasing validation batches before the validation program is complete, is permitted under Annex 15 only "in exceptional circumstances, where there is a strong benefit-risk ratio for the patient", justified, documented in the validation master plan and approved (5.16); the results and conclusion must be "formally documented and available to the Qualified Person prior to certification of the batch" (5.17). The FDA puts it more narrowly: concurrent release of PPQ batches "will be used rarely", for infrequently manufactured products such as orphan drugs, radiopharmaceuticals or drugs with a short half-life, and for medically necessary drugs in short supply, in coordination with the agency; conclusions about the process may only be drawn after the complete PPQ evaluation.

Changes, requalification and the question of revalidation

The validated state is a statement about a defined process, not a permanent condition. Annex 15 requires written procedures for changes to starting materials, the process, equipment, premises, batch size, design space "or any other change … that may affect product quality or reproducibility" (11.2). Quality risk management evaluates the impact on, among other things, "validation, regulatory status" and plans "any necessary process validation, verification or requalification efforts" (11.4); after implementation, the effectiveness of the change is evaluated (11.7). Equipment is requalified "at an appropriate frequency", and the period has to be justified (4.1, 4.2). A process change without a documented impact assessment therefore breaches 11.4, even if the original validation report is flawless.

ICH Q7 sets a counterweight here that is often overlooked: without significant change and with confirmed consistency of the process, "there is normally no need for revalidation" (12.60). Revalidation is thus the consequence of a change or a trend, not a calendar event. The validation master plan under Annex 15 sections 1.4 and 1.5 records exactly that strategy, including requalification, change control and deviation management; quality risk management is repeated when new knowledge emerges (1.7). For medical device manufacturers, process validation under ISO 13485:2016 clause 7.5.6 comes on top; EN ISO 13485:2016 with its amendments AC:2018 and A11:2021 is harmonized under the MDR.

What follows for practice

Five checkpoints follow directly from the texts. First: the validation master plan states the number of batches and justifies it, even when the number is three (Annex 15 sections 5.19, 5.20). Second: the terms PQ and PPQ are defined in the plan, so that a reader with an FDA background and a reader with an EU background read the same thing. Third: CPPs, CQAs and acceptance criteria are in the approved protocol before the first validation batch runs, and every later change is a documented deviation (2.7 to 2.9, 5.21; FDA IV.C.3). Fourth: ongoing process verification is a program with a protocol, statistical evaluation and a report in the product quality review, not a closing document (5.28 to 5.32; FDA IV.D). Fifth: every change goes through change control with an assessment of the validated status (11.4), and revalidation follows from a change or a trend, not from the calendar (ICH Q7 12.60).

Entourage supports validation strategy, IQ/OQ/PQ protocols and the setup of ongoing process verification in Process Validation, and, for new builds and conversions, the upstream qualification of facilities and utilities in GMP Facility Qualification & CQV.