Technical resources
Continued process verification for low-volume peptide campaigns
Low production volume makes process trends harder to interpret, but it does not remove the need to review them. Define comparable observations, retain campaign context and investigate directional changes before a release failure becomes the first signal.
A peptide process produces only a few campaigns each year. Every batch meets its release criteria, yet one processing time is gradually lengthening. The quality team needs to decide whether this is ordinary variation, an artefact of changing records or a reason to investigate. A continued process verification programme should make that decision possible.
FDA’s validation guidance treats continued verification as assurance that a pharmaceutical process remains controlled during routine production. The practical plan below is a proposed way to organize sparse campaign data. Its use on a research or cosmetic project does not establish a medicinal manufacturing status.
Define an observation before drawing a chart
Choose a small set of indicators with a reason for inclusion. Examples for discussion include an operation’s duration, recovery at a defined process boundary, or the frequency of unplanned interventions. The relevant indicator depends on the actual process; a universal dashboard can obscure more than it reveals.
State the denominator. A count of interventions per calendar month means little when the number of batches changes. Define whether the comparison uses batches, operating hours or another justified exposure measure. Keep the raw count available so a rate based on very little activity is recognizable.
Preserve equipment configuration, material version, analytical method and process revision beside each observation. If a timing field changes from manual recording to an automated timestamp, mark the change. The resulting step in a graph may arise from measurement practice.
Keep a small dataset honest
A run chart with dated, annotated observations can be more informative than an elaborate capability calculation supported by few data. Have a statistician choose appropriate tools and explain uncertainty. The NIST process-stability guidance distinguishes a stable process from one whose behavior changes over time, while its capability discussion relates stable process output to specifications.
Do not treat repeated measurements from one batch as a large history of independent batches. Keep within-batch and between-batch observations distinguishable. Where different products share equipment, pooling their results requires a justification based on the process question, not the desire to make a chart look populated.
Set review triggers with a response attached
Identify who reviews directional movement, repeated interventions or a change following maintenance. Statistical signals and specification failures need different descriptions. A result inside a release limit may still warrant investigation if the process pattern has changed.
For each proposed trigger, name the first useful evidence to inspect. A lengthening operation might lead the team to examine how its start and finish are defined, whether the equipment configuration changed, and which material attributes were recorded. This is an investigation route, not a diagnosis of any actual batch.
- Record the observation and why it triggered review.
- Check data comparability before assigning a process cause.
- State whether additional sampling or a technical study is needed.
- Keep the batch disposition decision under its existing approved procedure.
Review after inactivity as well as after production
Long gaps between campaigns create a practical review point. Before restarting, compare the intended conditions with the last supported process state. Confirm which equipment work, method revisions or personnel changes need assessment. A calendar-based review can document that no new production data exist without claiming that the absence of observations proves continued performance.
The monitoring plan should evolve when evidence grows. Preserve the rationale for changing chart limits or review frequency, and retain the earlier presentation so the change remains visible. Discuss these requirements through the CDMO process review as part of planning repeat campaigns.
Questions and answers
Can a small campaign history support a capability index?
Sometimes, but its reliability depends on the data and assumptions. Obtain statistical advice, report uncertainty and avoid treating a numerical index as strong evidence when process stability or sample independence remains unresolved.
Should batches with different process revisions share one trend line?
Only with a justified explanation. Mark revisions and assess comparability; separate presentations may be needed when the change alters the meaning or distribution of the measured attribute.
Why retain the count behind an intervention rate?
A rate based on one batch and a rate based on many batches can look identical while carrying different uncertainty. Keeping the count and exposure makes that limitation visible to the reviewer.
Does an in-specification trend require automatic batch rejection?
No. A trend signal calls for the review defined in the monitoring plan. Batch disposition follows its approved procedure, while the investigation assesses whether the process or its control strategy needs attention.
What can a periodic review say when no batches were made?
It can document the absence of new production evidence and assess changes during inactivity. It cannot convert an empty period into proof that process performance has remained unchanged.
