Flawed Process Validation, Ineffective Quality Unit Cited in Warning Letter

At the International GMP Conference held virtually in early March 2021 co-sponsored by the University of Georgia at Athens and FDA, FDA Office of Regulatory Affairs National Drug Expert Captain Ileana Barreto-Pettit presented FDA drug GMP warning letter trends and provided in-depth case studies from recent drug Good Manufacturing Practice (GMP) inspections illustrating agency concerns and findings.

Barreto-Pettit has been with FDA for 22 years.   She has been a drug investigator since 1999 and a Drug National Expert Investigator since 2017 and is a Captain in the U.S. Public Health Service Commissioned Corps.

[Related: Looking for 483 observations related to “process validation.” Register for a FREE 483 Observation Report using a specific keyword.]

The GMP inspection case studies Barreto-Pettit provided include an in-depth analysis of the findings, lessons learned, and how companies can avoid similar shortcomings.  Areas examined in the case studies are:

• The first country-wide import alert issued by FDA
•  Inadequate deviation investigation
• Inadequate product specifications and a product recall
• Process validation
• A different perspective on process validation and the culpability of the quality unit

The first four case studies were discussed in the first three parts of this story, available here, here, and here.  The fifth and final case study is examined in this final part of the story.

Case Study 5: Process Validation, Quality Unit Issues for Drug in Shortage

The fifth and final case study in Barreto-Pettit’s presentation was based on the findings from a mission-critical preapproval inspection for an injectable drug product that was in shortage.  This was the first commercial drug product for this firm at this location.  The product in the application was a sterile liquid drug product filled in large volume parenteral bags that are terminally sterilized.

At the time of the preapproval inspection, process validation had been completed.  On inspection, FDA found “a very highly automated and integrated manufacturing and filling process,” the national drug expert reported.  “And according to the quality unit, the process validation effort had been considered successful.  Therefore, we reviewed the process validation documents.”

The manufacturing process flow for the product begins with a compounding step, which in this case is a simple step in which a few ingredients are added to water.  That liquid is then pipe transferred into form-fill-seal machines.  The bags are formed and printed, filled, and sealed in these machines.

From there, they are sent to a conveyor through an auto tray loading and unloading into the sterilization chambers.  Then the bags go through a step to remove the moisture from the outer surface of the bags before going into automated leak detection equipment.

From there, the bags continue on a conveyor belt to a semi-automated bag inspection by visual inspectors.  Next, they go automatically to the next station, which is an over-wrap plastic bag that goes over the primary container to protect the outer surfaces and the ports.  And then from there to the final secondary packaging and labeling station. In reviewing the completed process validation data, FDA was provided with a table, which contained Process Performance Qualification (PPQ) information.  “As you can see, the firm manufactured four PPQ batches, one of which required rework due to issues with the over-wrap step crushing the ports of the bags during sealing.  When we first got this list, it did not have the last column—the yield column” (Figure 1).

Problematic Process Yield and Reconciliation Issues

What FDA investigators found concerning about the table was the number of units produced for batches that had a theoretical size of 30,000 liters.  For a yield of 100%, the company should have produced around 30,000 units of the 1,000 ml bag size and around 60,000 units of the 500 ml bags.

The table shows that the number of units produced was well below the expected outcome.  Instead of 30,000 units produced, for batch number three, there were only about 8,000 bags.  The same with the fourth batch, about 8,000 bags.  And for the 500 ml size, there were anywhere between about 20,000 and 26,000 bags as opposed to 60,000 bags.

The investigators asked for the calculated yields and were provided with those figures after they calculated them.  And as you can see, the total yield ranged from about 28% to 45%.  In addition, the reconciliation yield for each batch was also out of the normal range with values around 69% or so.  The company could not account for 30% of the batch.

“To make things even more difficult,” Barreto-Pettit commented, “they had not conducted investigations into these low theoretical versus actual yields or into the low reconciliation yields.  And the reason they had not conducted these investigations is that they had not set yield parameters in each manufacturing step or a total overall yield.”

At the time of the preapproval inspection, process validation had been completed

A primary observation on the FDA 483 had to do with inadequate process validation.  The low total batch yields and the lack of documentation for having such high rejection rates indicate that the process is not in control.

In addition, the regulations require that yield parameters be established for each manufacturing stage.   The company had not done that for the final or intermediate steps, including the compounding and form, fill and seal, and sterilization steps.

“Each step needs to have specific yield parameters, and if they do not meet those parameters then an investigation needs to be conducted,” she pointed out.  “They did not do that.  They did not have yields even for reconciliation to make sure to account for all the materials used for each batch.”

In addition, other information that should be in the batch record, such as extensive stoppages, interventions, or response to critical alarms, was not in the batch records.  As a result, it was “very difficult” to determine how well each batch ran in the processing line “because there were not many comments within the batch records.”

Inadequate Automated Systems and Quality Unit

During the inspection, investigators examined the electronic programming of the form-fill-seal equipment.  “We found that it had a lot of settings for critical alarms,” Barreto-Pettit commented.   “Since this was such an automated system that required little intervention by the operators, they had set alarms to go off so the operators would take action.  What we found was that some of these did not match the established process parameters.”

Some of the alarm settings were wider than the process parameters, therefore, they would not go off unless they were far out of range.  The company did not have an explanation for the wider settings.

For these and many other reasons in the 483, investigators listed observations for an inadequate quality control unit under 21 CFR 211.22(d).  The quality unit approved the flawed process validation, did not conduct investigations, and did not identify the deficiencies that were occurring in the process.

Process Validation Expectations

In her presentation, Barreto-Pettit emphasized some of the expectations for stage one process validation, which is also known as process design (Figure 2).

During this phase, FDA expects the company to build knowledge and understanding of the quality and interaction of raw materials as well as the process and equipment to determine the greatest sources of variation to establish a control strategy.  Manufacturers need to understand the functionality and limitations of the manufacturing equipment and how to detect the variation and its degree to implement procedures and strategies to maintain the process under control.

It is expected that the process is challenged.  Sometimes design of experiments (DoE) is used with different component lots, different operators, different environmental conditions as applicable, and the measurement systems in the production setting to identify the optimum parameters and conditions for the specific product.

Each step needs to have specific yield parameters

“In this case as evidenced by the low yields and the lack of information in the batch record, the company had a lot of rejected bags that were either filled or unfilled and multiple rejections from form-fill-seal to packaging,” she pointed out.

“Telling us that your company is overly conservative and it is rejecting more bags than needed does not mean, at least to me, that you understand the process or have confidence in the equipment measurement systems and the control strategy that you have implemented.”

“We did not feel during the inspection that there was sufficient knowledge about this equipment and the rejection rate at each station and the root causes for the rejections.  As I said before, there were no investigations as the quality unit did not question the low yields for the validation batches.”

Quality Unit Responsibilities

In closing, the National Drug Expert emphasized the responsibilities of the quality unit in the pharmaceutical quality system.

“As you know, the quality unit has overall responsibility for the quality and safety of the drug products that you manufacture and distribute to the US market.  If we find during inspections that companies are not identifying issues, are not following their responsibilities, or are not implementing systems to ensure that the products are consistently manufactured, those observations will appear in a 483 and could potentially be cited in a warning letter.”

“When it comes to contract manufacturers, I often find that the quality units do not have a robust oversight of these types of activities.  A company needs to find ways to make sure that whatever product is contract manufactured or tested at a third-party facility is following appropriate procedures and CGMPs and that the product is manufactured in accordance with CGMPs.  Make sure that you always have oversight of all operations throughout the life cycle of the product as well as the facilities and equipment.”

[Related: Looking for 483 observations related to “process validation.” Register for a FREE 483 Observation Report using a specific keyword.]