What Stage 3 is and what it is not

Process validation stage 3 is a monitoring program, not a repeat of qualification, and confusing the two is the most common error in commercial biologics manufacturing. Stage 2 process performance qualification answers a one-time question: does the process, run at commercial scale under enhanced sampling, perform as designed? Stage 3 answers a continuing one: is it still performing that way in batch 40, batch 100 and batch 300?

Three things process validation stage 3 is not. It is not release testing, which judges one batch, while continued process verification looks across batches for patterns no single result reveals. It is not the annual product review, a periodic document that summarizes CPV outputs. And it is not revalidation; where a trend shows the process has changed, CPV triggers investigation and, if needed, change control and comparability, not a re-run of PPQ.

The practical output of continued process verification is a set of control charts and capability indices for every critical parameter and attribute, updated as each batch is released, with defined rules for when a signal becomes an investigation.

Regulatory basis: FDA 2011 Stage 3 and EU Annex 15 ongoing process verification

Both major regulators require continued process verification under different names, and a biologic sold in both markets needs one program that satisfies both texts. The FDA's 2011 guidance Process Validation: General Principles and Practices defines Stage 3 as the stage in which ongoing assurance is gained during routine production that the process remains in a state of control. It calls for an ongoing program to collect and analyze product and process data, covering process trends and the quality of incoming materials, in-process material and finished product, statistically trended by trained personnel.

EU GMP Annex 15, Qualification and Validation, in operation since 1 October 2015, covers the same ground in sections 5.28 to 5.32 under "ongoing process verification during lifecycle". Manufacturers should monitor product quality to ensure a state of control is maintained throughout the lifecycle, using statistical tools where appropriate, and the extent and frequency of ongoing process verification should be reviewed periodically.

ICH Q10 (2008) frames both as elements of the pharmaceutical quality system: a process performance and product quality monitoring system that supplies the data for continual improvement. In the United States, the annual product review under 21 CFR 211.180(e) is the compliance anchor the CPV program feeds.

Building the CPV plan: parameters, sampling, statistical tools

A continued process verification plan is a controlled document approved before the first commercial batch, and it must answer three questions for every monitored variable: what is measured, how often, and what statistical rule turns a value into a signal. The parameter list comes from process characterization and the PPQ report.

Two statistical points matter. Control limits are calculated from process data, typically at three standard deviations around the mean, and are not specification limits; a parameter can be inside specification and still out of statistical control. Capability indices such as Ppk compare process spread with the specification range; a Ppk below about 1.0 means out-of-specification results will occur at a predictable rate even if every batch so far has passed. Many programs set Ppk 1.33 as the threshold for reduced sampling.

Review cadence and links to APR/PQR

Continued process verification works on three clocks, and a plan that defines only the annual one is not a monitoring program. The first is per batch: as release data arrive, charts are updated and signal rules checked. The second is periodic, commonly quarterly, when a CPV report reviews all charts, recalculates control limits if justified, and documents trends and their disposition. The third is annual, when CPV data are consolidated into the annual product review or product quality review.

The annual review has its own regulatory content. Under 21 CFR 211.180(e) it evaluates each product's quality standards at least annually to determine whether specifications or procedures need to change, including complaints, recalls, returns and investigations. EU GMP Chapter 1 requires a product quality review covering starting materials, in-process controls, out-of-specification batches, deviations, changes, stability, complaints, recalls and previous corrective actions. CPV supplies the statistical backbone for the first three items.

Nothing in the annual review should surprise anyone; a trend that first appears there means the per-batch and periodic clocks failed.

Handling trends, deviations and post-approval changes (ICH Q12)

A continued process verification signal is not a deviation, and treating every statistical alert as a quality event buries real problems in paperwork. The plan should define a graded response: a single run-rule violation is documented and the next batches watched; a confirmed trend, such as three consecutive batches with rising aggregates while still in specification, opens an out-of-trend investigation; and any value outside specification follows the out-of-specification procedure.

Investigations in commercial manufacturing usually trace trends to a short list of causes: a new raw material lot, resin or membrane aging, instrument drift within calibration tolerance, or an unrecorded change in operator practice. Correlating performance-attribute charts with raw material lot data finds most of them.

When the cause requires a process change, the regulatory path depends on how the change was anticipated. ICH Q12 (2019) introduces established conditions, the elements of the process and controls that are binding in the marketing authorization, and the post-approval change management protocol, which pre-agrees the data required for a defined change. A sponsor with defined established conditions can implement many CPV-driven improvements under a lower reporting category; changes to established conditions still need a supplement or variation and, for biologics, an ICH Q5E comparability exercise.

Dividing CPV responsibilities between sponsor and CDMO

When commercial manufacturing is outsourced, continued process verification is a shared program, and the quality agreement must say who does what, because regulators hold the marketing authorization holder accountable while the site holds the data. The sponsor retains final responsibility in every row of the table.

Sponsors evaluating a CDMO for commercial manufacturing should ask for an anonymized CPV report and the site's statistical procedures. A site that trends only release results, with no process parameter data, cannot support process validation stage 3.