Draft Guidance: Considerations for Complying With 21 CFR 211.110 Guidance for Industry

Thank you for sharing this recommendation! It's great to see how the Draft Guidance builds on the foundational concepts from the 2004 PAT Guidance and ICH Q8 (R2), while also integrating with ICH Q10.

I'm curious to hear from others in the forum: How do you think the adoption of in-line and on-line sampling will impact your current practices? Are there any challenges or benefits you've experienced with these methods? Let's discuss!

The guidance sets the guardrails but certainly leaves ample room for adopting the proper in-line and on-line sampling technology and protocols.  It would be very valuable too hear from anyone that is currently undertaking or perhaps has already completed in-line or on-line sampling programs for a CM process.  Broad details that do not break confidentiality would be very beneficial to initiate a discussion.  I am particularly curious s to what are the biggest challenges being encountered .  

Stephen, thank you for bringing up this excellent topic. I’d like to share my perspective based on my experience with continuous manufacturing (CM) in solid dosage forms.

From a solid dosage standpoint, the key decision is often between in‐line and at‐line monitoring. In other CM processes—as you mentioned—the strategic decision typically focuses on in‐line versus on‐line sampling. Based on my background in solid dosage, I will clarify these differences from my perspective.

When choosing between these options, the Development, Quality Assurance (QA), and Business teams must clearly define the purpose of the technology—whether it is for monitoring, control, or real-time release testing (RTRt). While it’s possible to implement either approach or a combination of both, each option has its own advantages and disadvantages.

At-line:
This approach is primarily employed at the end of a batch or continuous process, serving as a final checkpoint. System corrections are limited—except in specific cases (such as loss-on-drying in granulation) where additional time permits adjustments. However, an at-line test for API concentration prediction conducted post the compression stage does not allow for any correction; at that stage, any non-conformance drug product must be manually segregated unless the line design includes a post hold-up stage that allows the material to be evaluated prior moving to the next stage: coating.   
It is important to note that at-line measurement frequency is significantly lower than that of in-line monitoring which could have an impact on process yield. Nevertheless, depending on factors such as API interference with other excipients, chemical characteristics, concentration, and additional process and instrumentation considerations, an at-line control may be the only viable option.

In-line:
This one will be evaluated for control purposes. The in-line monitoring approach is inherently more complex. Initially, in-line monitoring was envisioned immediately after the blender; however, advances in technologies and process understanding now permit its implementation within a press feed frame. When positioning an in-line Process Analytical Technology (PAT), consider the following factors:

• Dynamics and Model Interference:
  Both the feed frame and the blender output exhibit unique dynamic characteristics that must be taken into account when assessing model performance, particularly for control purposes.
• Residence Time Distribution (RTD) vs. Yield:
  When positioning the in-line PAT in the feed frame, throughput becomes critical. Can the control system measure, evaluate, and respond quickly enough to reject nonconforming material? Furthermore, if in-line monitoring is implemented at the blender outlet, would an additional at-line PAT testing point, post-press, would be required to confirm blend homogeneity? Take in mind that equivalency study can get quite complex (time, effort, expertise and material availability) to proof the previous assumption.
  Note: These are not the only options or combinations to implement either at-line, in-line or combinations. There are multiples ways to do this. It depends on case-by-case. 
• Instrument Resolution vs. Scan Rate:
  It is crucial that the instrument in the feed frame possesses sufficient resolution to capture a high-quality spectrum necessary for developing a robust chemometric model. The resolution is dependent of multiple factors (ex. integration time or scans and co-ads or number of average spectra). This can have an influence on the time that the instrument takes to generate a raw spectra without to mention the control strategy logic post the prediction. Therefore, always keep in mind the RTD and the inherent delays in measurement, data transfer, evaluation, and control actions to ensure your system is capable to divert, on time, any non-conformance material.

There is additional aspect to consider but it will depend on the process and drug product risk assessment. I’d welcome insights from others, not only regarding solid dosage but also in drug substance manufacturing.

This article may be of interest for those following along with this discussion: What FDA Draft Guidance Tells Us About In-Process Control Strategies. It's a guest column and the author also touches on in-line sampling.

This article may be of interest for those following along with this discussion: What FDA Draft Guidance Tells Us About In-Process Control Strategies. It's a guest column and the author also touches on in-line sampling.