A Case Study In Continuous Process Verification

By Irwin Hirsh, Q-Specialists AB

Continued process verification (CPV) can be used to evaluate business-driven optimization of downstream purification conditions that remain fully within the registered design space.

To illustrate what I mean, here’s a hypothetical scenario in which a downstream development lab sought to improve yield and column-loading efficiency.

The Setup

A bacterial recombinant protein process was producing excellent upstream harvests, yet downstream yield was limited by overly restrictive peak cutting parameters in the first chromatography column. The solution: operate at the lower end of the filed column load range to broaden fractionation and recover more product without compromising HCP clearance or overall quality.

Manufacturing Context

Expression system: Recombinant E. coli producing a therapeutic protein.

Issue observed: Downstream chromatography (Column 1) showed increased variability in HCP clearance and final purity despite consistent upstream harvests.

Prior adjustment: The purification team had narrowed the Column 1 fractionation window to better remove HCP, which unfortunately also reduced overall yield.

Proposed Optimization
Load adjustment: Operate Column 1 at the lower end of the registered load range

Hypothesis: Reduced load would:

1. improve fractionation resolution,
2. reduce HCP carryover, and
3. increase final yield by reducing product loss from narrow peak cuts.

Status: Small-scale models support the hypothesis. The range is already filed. No change control is required. CPV is used to verify and build understanding.

Real-World Considerations

Running at the lower end of the column load range introduces trade-offs that must be managed:

Increased cycle time: Lower mass per load may require additional column cycles to meet production targets, extending overall process duration.

Capacity constraints: Longer cycles may strain scheduling or facility throughput, especially for multi-product lines.

Operational complexity: Additional cycles introduce more opportunities for deviations, buffer prep, and equipment wear.

Buffer consumption: More column volumes can lead to higher consumption of costly buffers and cleaning agents.

Why Use CPV for Movement in the Filed Range?

We have to acknowledge that the data generated in clinical production and demonstration batches is indicative but not conclusive (statistically speaking) and thus we maintain our confidence in the validated state with such CPV plans in combination with the routine QC work used for batch release.

Including this change in loading as part of the CPV will provide data both critical-to-quality and critical-to-business. A deep-dive into CPV run data:

will tell operations:

• if yield gains and robustness outweigh operational burden
• confirm that yield and purity outcomes are sustainable in routine mfg.

this same data also helps:

• provide the verification of the validated state as runs at an extreme add to the process knowledge and confidence in its robustness.

As this optimization sits within the validated design space and requires no regulatory variation or change control. Still, a structured plan is essential to understand subtle process variations that affect downstream performance, and build deeper life cycle knowledge.

The CPV plan is not questioning the validity of the registered ranges but rather is using advanced analytics to ensure that optimization remains robust across routine manufacturing conditions and use this information as part of the formal commitment to the customer and health authorities for ongoing process verification and continuous improvement.

CPV Parameters Monitored

Table 1 CPV Monitoring

Type	Parameter	Purpose & Justification
CPP	Column 1 Load Volume (L/kg)	Primary adjusted input; hypothesized to improve yield
CPP	Fractionation Cut Window (A280)	Tightened previously; monitored for HCP separation and yield impact
CPP	Relative Retention Times	Looking for the peak of the peak and overall chromatogram properties around elution to be consistent across all batches
CQA	HCP (Post-Column 1, Pre-UF/DF)	Key driver of purity and reprocessing risk
CQA	Final Product Purity (SEC/UV)	Product quality attribute influenced by load/fraction
CQA	Aggregates	Sensitive to stress or overload on Column 1
IPC	Pool Yield Post-Column 1	Critical to business outcome; affected by peak cutting
IPC	Final UF/DF Yield	Confirms end-to-end benefit of new load strategy
Context	Resin Age, Buffer Lot, Harvest OD	Normalize and investigate multivariate contributors
Upstream	Harvest OD600, Total Biomass, HCP Load	Characterize harvest variability and its downstream impact
Upstream	Soluble Protein Yield, Host DNA Levels	Further classify harvest quality attributes

Advanced Analytical Techniques In CPV

Unlike standard QC trending, which often relies on univariate control charts and basic rule-based signals (e.g. Nelson Rules 1–4), CPV must often employ deeper analyses such as multivariate data analysis (MVDA) like PCA, PLS, and clustering approaches to reveal process insights:

• detect hidden correlations (e.g., between upstream OD, load volume, and impurity retention),
• visualize shifts or batch clusters that precede downstream purity changes,
• distinguish variability from noise in complex systems,

• understand how co-variability among process parameters (e.g., pH, conductivity, and load volume) affects final product outcomes.

Why Development Might Miss These Patterns:

• Development studies are often run under narrow experimental conditions with little batch variability.
• Small n-values in development limit true MVDA capability.
• Focus is often on defining ranges, not deeply characterizing variability in routine conditions.
• Even DOE studies in development suffer from gaps such as unit operations studies being independent from the whole process or variation being significantly different at commercial scale than experience at small scale.

Holistic Process View:

• CPV integrates upstream variability (biomass quality, harvest timing) into downstream control context.
• It evaluates how cumulative variation propagates through purification steps.
• This work is done at commercial scale on commercial product. Designing in an adequate sample size is not an issue in that it does not impact time to market
• Little to no patient risk. The risk is entirely related to COGS for the producer.

Business Risk Mitigation And Regulatory Alignment

• These CPV plans are explicitly framed as life cycle process understanding within validated ranges.
• No suggestion is made that current controls are inadequate.
• Results of the CPV work are “paid for” by the productivity increase
• Changes, even if they do not become routine, add to the body of evidence for a well controlled process and can become part of the APR/PQR

CPV Is Not QC Trending

• QC trending is compliance-driven and ensures batch-by-batch control using predefined specification limits.
• CPV is knowledge-driven and monitors the state of control over time to detect emerging risks and identify opportunities for improvement.
• CPV is not a replacement for QC trending but rather a complementary system with a broader process understanding objective.

Lessons For The Industry

CPV is yet another regulatory expectation that can also be a proactive tool for driving business-beneficial improvements (see here for how QRM helps speed up time to market).

When applied correctly, CPV strengthens the bridge between compliance and continuous improvement. This case exemplifies optimization within filings through structured, data-driven life cycle monitoring. Advanced data science complements routine QC by offering a system-level understanding of process control.

About The Author:

Irwin Hirsh has nearly 30 years of pharma experience with a background in CMC encompassing discovery, development, manufacturing, quality systems, QRM, and process validation. In 2008, Irwin joined Novo Nordisk, focusing on quality roles and spearheading initiatives related to QRM and life cycle approaches to validation. Subsequently, he transitioned to the Merck (DE) Healthcare division, where he held director roles within the biosimilars and biopharma business units. In 2018, he became a consultant concentrating on enhancing business efficiency and effectiveness. His primary focus involves building process-oriented systems within CMC and quality departments along with implementing digital tools for knowledge management and sharing.