Written by Tracy Hibbs and Tony Sacchetti, Waters Corporation

Artificial intelligence (AI) is poised to take routine work and shift it to fully automated workflows. Currently, 80% of AI projects fail.¹ As with most paradigm-shifting technology, it will require time, patience, and strong data governance to realize the potential gains. It requires deep expertise to ensure the AI performs as expected for your intended use and vigilant, continuous oversight to reduce the potential for risks to be realized.

Failure is not an option

With approximately 95% of pharmaceutical companies investing in AI,² it is clear hopes are high that it will accelerate drug development, secure supply chains, increase operational resilience, streamline audit predictability, and reduce time and cost to patients. At the same time, regulatory agencies are investing to increase the speed of filing response, identify potential supply chain issues and trends, and pinpoint where to focus inspections. Innovation and regulation are happening in parallel, increasing the criticality of getting it right, or risk falling behind.

For a closer look at the regulatory landscape and at inspection readiness specifically, see our recent post, “Is Your Organization Ready for AI Inspection?”

“In pharma, 75 to 85 percent of workflows contain tasks that could be enhanced or automated by agents, potentially freeing up 25 to 40 percent of an organization’s capacity.” ³

Considering the extensive investments that have been made, organizations are still experiencing a lack of ROI. Whether it is efforts to build homegrown AI with existing resources, gaps in needed expertise, siloed and disparate activities, or data governance issues impacting success is to be determined, and in many cases, it is likely combinations of challenging dynamics.

From a practical standpoint, there are critical areas, such as AI validation and GMP compliance, where things can break down:

You cannot AI your way out of bad data

Data integrity is not optional and ALCOA++ is essentially the regulatory floor. Any potential data integrity issues that exist in your input data will persist in your AI—it inherits and amplifies your data. You may miss issues if your approach is black box and you lack a clear understanding of how the model you use works and where it fails. Your data governance and validation practices must be in a continuous state, especially if you are attempting to adopt adaptive AI.

These fundamentals must be working in concert to truly realize the potential of AI.

If you consider agentic AI, for example, it is not just AI. It is AI that plans, reasons, and executes multi-step actions toward a goal—language that mirrors the FDA’s own working definition of agentic AI.⁴ It is not generating one single, simple prediction, it is making a sequence of decisions, often with autonomous correction. Think of an AI that does more than just flag an OOS result, it autonomously launches the investigation, retrieves prior batch records, drafts the deviation report, and routes it for approval.

Your AI vendor choice becomes your AI governance choice

How will you ensure your AI consistently produces reliable outcomes? If you are building solutions for your organization, your custom code carries a high validation burden compared to partnering with an organization that has intentionally designed and built solutions for high-value use cases. You also gain the burden of building and maintaining all infrastructure required to sustain the custom code you developed.

AI accelerates transformation at warp speed, but only mature organizations can safely control and sustain that speed. As your organization looks to accelerate, your vendor selection becomes a critical piece that is essentially a regulatory decision.

Things to consider:

• Choose vendors for governance posture, not merely feature breadth.
• Look to get it right the first time over getting there first.
• Execution that increases confidence is as critical as innovation.

The right vendor will ensure there is AI transparency, with documented inputs, outputs, decision boundaries, and limitations. Alongside this, vendor assessments against regulations, data flow and process flow maps, and risk assessments are imperative to streamline your compliance efforts and support your quality risk management. It is critical to have a clear understanding for your quality unit (QU) and frameworks that provide explicit information for human oversight, locations of where the human can be put into the loop, and authorization.

Getting it right means continuous, embedded validation with performance monitoring, drift detection, and change control embedded in the same platform of capabilities versus an external bolt-on or afterthought.

Is your organization ready to leap ahead?

Connect with a Waters expert. Bring your gaps and we will bring the compliance expertise needed. Together, we will keep your data inspection ready.

References

1. Rand^®, Gartner^®, Stephen Ferrell – CPO Valkit.ai, ISPE^® GAMP^®
2. Mordor Intelligence (2025)
3. Reimagining life science enterprises with agentic AI, McKinsey (2025)
4. FDA Expands Artificial Intelligence Capabilities with Agentic AI Deployment (December 2025)

Written by Tracy Hibbs and Tony Sacchetti, Waters Corporation

Exploring artificial intelligence (AI) and machine learning (ML) in regulated laboratories governed by Current Good Manufacturing Practices (CGMP) may be new for your organization. Inspecting AI is not new to regulators. If we look at FDA, for example, use of AI/ML-enabled medical device authorizations have grown sharply since 2018, as shown in the chart below.^1,2

Regulatory expectations: The fundamentals remain

As an industry, we have discussed AI regulation for years, while regulators have moved quickly to address rapid technological change. The fundamentals remain unchanged: data integrity, data governance, and risk management continue to underpin AI oversight, and inspectors are placing even greater scrutiny on these core principles.

The foundational guidance used daily and incorporated into your Pharmaceutical Quality System (PQS) remains unchanged. ICH Q9(R1) Quality Risk Management³ and PIC/S Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments⁴ continue to anchor responsible use. Newer papers and draft guidance extend that foundation specifically to AI, including EMA Annex 22⁵ and the FDA and EMA jointly issued Guiding Principles of Good AI Practice in Drug Development.⁶

Between them, there is more than enough guidance to inform a responsible AI adoption strategy. For a wider view of the regulatory landscape, read our blog, “AI in GMP: Navigating Evolving Expectations for Regulated Labs.”

Enforcement is not theoretical

Inspections and authorizations to date have leveraged existing regulations and finalized guidance. Regulatory actions have leveraged the same. The FDA published multiple warning letters referencing AI from 2023 to 2026, and an April 2026 CGMP warning letter was the first to cite inappropriate use of AI as a stand-alone CGMP deficiency.⁷

Citing 21 CFR 211.22(c) ⁸ and the responsibilities of the quality control unit, the warning letter message is direct: if AI is used to support CGMP activities, recommendations and outputs from AI agents must be reviewed and cleared by an authorized human representative of your quality unit (QU).

Quality unit responsibility is not reduced when AI is involved

Accountability remains unchanged. Leveraging AI assistance does not reduce the obligation of your qualified personnel. It is critical that your QU knows where AI is leveraged across your organization. If this visibility does not exist today, building it should be an immediate priority.

Your supply chain partners are also in scope. Contract development and manufacturing organizations (CDMOs) and contract testing labs operate as extensions of the manufacturer, and inspectors continue to treat them that way. If a contract organization uses AI to draft batch records, specifications, or SOPs without governance, the risk transfers to the sponsor. Quality agreements and supplier audits now need to address AI use explicitly: permitted applications, human oversight requirements, audit rights, and evidence expectations.

Your AI is rated whether you have formally rated it or not

The draft FDA⁹ framework provides a practical way to assess regulatory burden. The complexity of the AI system and its impact on product quality define the category, the intersection signals the oversight required. This principle is consistent across guidance: oversight must be commensurate with risk to product quality and patient safety.

Looking at risk categories, expectations vary:

An unclassified AI system is not an unregulated one. Inspectors will assign a category whether your organization has or not, and they will expect controls and evidence to match.

There are expectations in your inspections

If an inspector were to audit you tomorrow, could you answer each of these five questions with confidence and documented evidence?

1. Where is AI used in your CGMP/GxP workflows today, and at what risk tier?
2. Who reviews and approves AI-generated outputs, and what is the documented evidence?
3. How is your training data separated from your validation and test data?
4. How do you detect data drift and model performance degradation in production?
5. What is your AI literacy plan and your agentic AI governance for staff?

Where does a state of control break down?

Demonstrating a state of control is essential in regulated labs. With AI in use, the demonstration could potentially falter in several predictable ways. Each of the patterns below maps to one of the five inspector questions above:

The time to be ready is now

Now is the moment to assess your readiness for AI adoption under evolving global regulations. Whether you are exploring AI for the first time or scaling existing capabilities, alignment with regulatory expectations unlocks AI’s full potential, without compromising product quality or patient safety.

Connect with one of our experts to help assess your readiness for AI and for AI inspections.

References

1. Innolitics 510(k) Year-in-Review (2025)
2. Imaging Wire · FDA AI/ML Device list (December 2025)
3. ICH Q9(R1) Quality Risk Management
4. PIC/S PI 041-1, Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments
5. EMA Annex 22 (Draft 2025)
6. FDA and EMA Guiding Principles of Good AI Practice in Drug Development (January 2026)
7. FDA Warning Letter 722591 (April 2026)
8. United States Code of Federal Regulations 21 CFR 211.22
9. FDA Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products, AI/ML Risk Classification Framework (Draft January 2025)

Additional Resources

• ISPE^® GAMP^® Guide: Artificial Intelligence (July 2025)
• European Union, AI Act (Regulation (EU) 2024/1689), with Article 4 AI literacy obligation effective August 2, 2026

Written by Tracy Hibbs and Tony Sacchetti, Waters Corporation

Regulators are already inspecting AI use, how prepared is your organization?

For years, the pharmaceutical industry has anticipated regulatory clarity on the use of artificial intelligence (AI) and machine learning (ML) in regulated labs. This clarity has emerged, not only through guidance, but also through enforcement. For quality control (QC) labs, this moment signals both opportunity and accountability. AI can deliver speed, consistency, and efficiency, but only with strong AI governance and when deployed under conditions that preserve transparency, traceability, and scientific integrity.

Foundations: The regulatory path that got us here

In 2021, the Danish Medicines Agency (DKMA) became one of the first European regulators to address AI in pharmaceutical quality systems when they took a bold step toward defining the role of AI in regulated environments by publishing a draft guidance entitled “Suggested Criteria for Using AI/ML Algorithms in GxP.”¹

Their work informed reflection papers and draft guidance from the European Medicines Agency (EMA) and the U.S. Food and Drug Administration (FDA), all formally recognizing AI and ML as viable tools in regulated pharmaceutical operations, including QC. These documents do more than permit AI—they define the conditions under which it can be trusted.

2021 DKMA Suggested Criteria for Using AI/ML Algorithms in GxP emphasizes:

• Use of static, supervised models for critical functions
• Independent test data and bias mitigation
• Strong data integrity and validation practices

2023 EMA Reflection Paper on the Use of Artificial Intelligence (AI) in the Medicinal Product Lifecycle² emphasizes:

• Transparency and explainability
• Risk-based validation
• Human oversight

2025 EMA with Pharmaceutical Inspection Convention and Pharmaceutical Inspection Co-operation Scheme (PIC/S) Draft Guidance Annex 22: Artificial Intelligence³ explicitly excludes adaptive or generative AI from critical GMP applications and mandates:

• Static, deterministic models only
• Locked training data and independent test sets
• Explainability tools (e.g., SHAP, LIME) and confidence scoring
• Defined operator roles in human-in-the-loop (HITL) systems

2025 FDA Draft Guidance Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products⁴ introduces a risk-based credibility assessment framework, focused on:

• Model transparency and traceability
• Data governance and independence of training/test sets
• Lifecycle management of AI models
• Context-of-use validation

Ten principles, two regulators, and one path

The FDA and EMA jointly published Guiding Principles for Good AI Practice in Drug Development.⁵ The synchronized framework is significant, indicating a renewed alignment between two influential regulatory agencies and signaling that organizations operating across jurisdictions can plan against a stable target.

The joint announcement highlights 10 critical areas:

Apply critical thinking

Core skills developed across your organization, from risk management to data governance and vigilant monitoring, become more critical when leveraging AI. Your organization remains accountable for your quality.

• AI reliability: Data quality is foundational, requiring strict data segregation, traceability, and version control. Lifecycle management is a cornerstone, as are data integrity and bias mitigation.
• AI successful implementation: Collaboration is an imperative requiring cross-functional teams including your analysts, QA, IT, AI ethics team, data integrity officer, and regulatory affairs team.
• AI decision impact: Understanding the criticality of the decision is not optional. Risk assessment good practices and data flow and workflow maps identifying your intended use underpin your implementation.
• AI oversight: Knowing the number of decisions being made and applying vigilant monitoring at each decision point are fundamental.

Why this matters: The risk to product quality and patient safety

The stakes in pharmaceutical QC are uniquely high. Every decision in QC labs has downstream implications for product quality, compliance, and, ultimately, patient safety. AI systems that are opaque, unvalidated, or poorly governed have the potential to introduce detrimental risks, including:

• Undetected anomalies in chromatographic data could allow substandard or contaminated products to reach patients.
• Adaptive models that change behavior over time may produce inconsistent results, undermining batch release decisions.
• Lack of explainability can obscure the rationale behind critical decisions, making root cause analysis and regulatory inspections more difficult.
• Insufficient validation or biased training data can lead to systematic errors, disproportionately affecting certain product types or patient populations.

This is why regulators are drawing a clear line. AI must not compromise the scientific rigor, traceability, or reproducibility that underpin GMP. Instead, it must enhance them.

What does this mean for QC labs?

For QC labs, this is a pivotal moment. AI/ML can now be deployed to:

But adoption must be deliberate. Labs and quality teams must ensure:

• Models are locked and validated before use.
• Outputs are interpretable and reviewable.
• Human reviewers remain accountable and trained.
• Strong quality oversight persists.

This is not about replacing analysts. It’s about amplifying your deep expertise and focusing attention where it matters most.

Looking ahead: The message is straightforward

Regulatory direction is clear: AI use in GMP areas is permitted, provided it operates within a framework of transparency, reproducibility, and oversight. As regulators continue to refine expectations, particularly around validation standards, model updates, and governance of agentic systems along with real-time monitoring, QC labs that invest now in compliance-enabled AI will be better positioned for the future.

Is your lab ready for the AI era?

Connect with our experts to explore how Waters is enabling labs to take the next step toward intelligent, always audit-ready systems.

References

1. Danish Medicines Agency (DKMA), Suggested Criteria for Using AI/ML Algorithms in GxP (2021).
2. EMA Reflection paper on the use of Artificial Intelligence (AI) in the medicinal product lifecycle (Draft 2023, Final 2024).
3. EMA GMP Annex 22 (Draft 2025).
4. FDA Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products (Draft January 2025).
5. FDA and EMA Guiding Principles of Good AI Practice in Drug Development (January 2026).

Additional Resources

• ISPE® GAMP® Guide: Artificial Intelligence (July 2025)
• FDA Computer Software Assurance (CSA) Guidance (February 2026)
• ICH Q9(R1) Quality Risk Management
• PIC/S PI 041-1, Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments

While investment decisions often focus on mass spectrometry (MS) detection, experience across clinical laboratories shows that chromatography is frequently the limiting factor, both scientifically and financially.

When chromatographic performance is unstable, the cost is not confined to poorer data quality. It appears as lost productivity, reanalysis, delayed reporting, and increased operational risk. In this environment, Ultra High Performance Liquid Chromatography (UHPLC) technology is essential for protecting return on investment (ROI) in clinical LC‑MS workflows.

The hidden financial cost of unstable chromatography

In routine clinical laboratories, chromatography must perform reliably across:

• Hundreds to thousands of samples per day
• Complex and variable biological matrices
• Multiple operators and long assay lifetimes
• Regulated and audited workflows

When it does not, the consequences are immediate and costly. Retention time instability and poor chromatographic reproducibility lead to:

• Wider MS acquisition windows
• Reduced dwell time and fewer data points per peak
• Increased risk of missed detections in large analyte panels
• Higher rates of manual data review, reruns, and batch failures

Each of these outcomes consumes analyst time, instrument capacity, reagents, and consumables; eroding throughput and increasing cost per result. Over time, these inefficiencies quietly undermine the financial viability of LC‑MS testing.

Retention time stability drives both performance and productivity

Retention time stability is often discussed as a technical metric, but in clinical LC‑MS it has direct economic implications. Stable retention times allow laboratories to:

• Use narrow, confident MS acquisition windows
• Increase dwell time per transition
• Generate more robust quantitative data per injection
• Reduce uncertainty that forces manual review

When retention times drift, acquisition windows are often widened “just in case.” This precaution reduces MS efficiency and sensitivity, especially for low‑abundance analytes, increasing the likelihood of reanalysis or inconclusive results.

In high‑throughput environments, retention time variability is one of the fastest ways to increase your cost per sample while reducing throughput.

UHPLC as a cost‑containment and risk‑reduction strategy

For clinical laboratories, UHPLC is not simply about achieving better separation—it is about controlling operational and financial risk. Robust, stable UHPLC reduces:

• Missed detections that require repeat analysis
• Batch failures triggered by chromatographic drift
• Analyst time spent on troubleshooting and manual review
• Variability that complicates trending and audit investigations

By stabilizing chromatography, laboratories stabilize workflow efficiency, instrument utilization, and reporting timelines—directly contributing to improved ROI over the assay lifecycle.

ACQUITY UPLC I‑Class PLUS System: Designed for long‑term ROI in clinical labs

The Waters ACQUITY UPLC I‑Class PLUS System was developed with decades of expertise. Rather than prioritizing maximum pressure or speed alone, the system is engineered to deliver:

• Highly stable solvent delivery for consistent retention times
• Precision injection supporting reproducible LC‑MS acquisition
• Robust operation suitable for continuous, high‑throughput use
• Design principles aligned with regulated clinical environments

The result is not simply higher performance, but fewer workflow interruptions, lower rework rates, and more predictable assay behaviour over time.

The takeaway

As clinical LC‑MS testing expands, laboratories must evaluate technology choices not only on analytical capability, but on total operational impact.

Unstable chromatography increases cost per sample, consumes valuable staff time, and introduces unnecessary risk. Stable, clinically optimized UPLC enables higher throughput, consistent performance, and predictable outcomes across the life of the assay.

With systems such as the ACQUITY UPLC I‑Class PLUS System, UHPLC helps ensure that chromatography strengthens the return on investment in modern clinical LC‑MS.

Explore how UPLC can reduce risk and increase efficiency in your clinical lab.

Chromatography is often the root cause of the variability observed at the mass spectrometer.

Today’s MS platforms are technologically mature, electrically stable, and highly reproducible over time. When signal changes occur, the detector is usually doing exactly what it was designed to do:  reporting the conditions under which analytes entered the ion source.

Those conditions are set upstream by chromatography.

Chromatography determines what is delivered to the MS

Before an analyte ever reaches the mass spectrometer, chromatography determines when it arrives, how concentrated it is at any given moment, and which other components arrive alongside it. Retention time, peak width, peak shape, and co‑elution patterns all originate in the LC system.

Any variability in these parameters is immediately translated into variability at the MS. From the detector’s perspective, changing chromatography appears as changing signal response, even if the MS itself is perfectly stable. This is why apparent “MS variability” so often persists despite tuning, cleaning, or servicing the mass spectrometer.

How chromatographic instability manifests as MS variability

One of the most common contributors is retention time instability. When retention times shift, acquisition windows must be widened to avoid missed detections. Wider windows reduce dwell time per transition, decrease the number of data points across the peak, and increase exposure to interfering signals. The result is poorer precision and less consistent quantitation.

Peak broadening, often caused by extra‑column dispersion, has a similar impact. When an analyte band is spread out after leaving the column, the same amount of analyte is delivered to the ion source over a longer period. Peak height drops, signal‑to‑noise suffers, and integration becomes less consistent. Again, the MS registers variability that originated from the chromatography.

Matrix co‑elution further compounds the problem. Inadequate chromatographic separation allows more background components into the ion source, increasing ion suppression or enhancement. These matrix‑driven effects are highly variable and can make MS response appear unstable, even though the underlying cause is insufficient chromatographic control.

Why does this matter more in clinical laboratories

Clinical LC‑MS workflows amplify the impact of chromatographic variability. High‑throughput operation, large multi‑analyte panels, complex biological matrices, and long assay lifetimes leave little tolerance for drift or inconsistency.

Small chromatographic changes that might be manageable in research settings become serious operational issues in clinical labs. Over time, they lead to increased manual data review, more frequent reruns or failed batches, difficulty trending system suitability metrics, and reduced confidence during audits and inspections.

Crucially, the mass spectrometer cannot correct for unstable chromatography. It can only report the consequences.

Stabilizing chromatography stabilizes the MS

When chromatography is stable, MS analytical variability decreases.

Stable retention times allow narrow, confident acquisition windows. Consistent peak shapes ensure reproducible ionization conditions. Reduced dispersion preserves peak height and signal quality. Together, these factors lead to more reproducible MS signals, improved quantitative precision, and more reliable detection of low‑level analytes.

In many cases, improving chromatographic stability is the most effective and sustainable way to improve overall LC‑MS performance.

The Role of UHPLC and Low‑Dispersion System Design

Modern UHPLC systems designed for clinical use address key chromatographic drivers of MS variability by delivering high efficiency alongside long‑term reproducibility.

Low‑dispersion system designs are particularly important, as they preserve the narrow peaks generated by UHPLC columns and maintain predictable retention time behavior.

By minimizing variability in analyte delivery to the ion source, these systems reduce apparent MS variability at its source.

The Waters ACQUITY UPLC I‑Class PLUS IVD System reflect this philosophy. The system design prioritizes stable solvent delivery, precision injection, ultra‑low dispersion, and consistent chromatographic behaviour over extended routine operation—attributes that directly translate into more reproducible MS data in clinical workflows.

The takeaway

When LC‑MS variability appears, it is tempting to look downstream. But in clinical workflows, the mass spectrometer is often responding correctly to unstable conditions created upstream.

Chromatography determines when, how, and under what conditions analytes reach the MS. If chromatography varies, then MS data will vary with it.

For clinical laboratories seeking reliable, scalable, and defensible LC‑MS, the most effective place to reduce variability is not the detector, but the chromatography that feeds it.

Learn more about the impact of UPLC on clinical analysis.

In pharmaceutical quality control (QC), consistency should mean more than an analyst ending a shift relieved that the results were repeatable and there were no obvious failures. It should mean having enough confidence in the process, the controls, and the data to know that those results are genuinely reliable, explainable, and sustainable over time. This is where Stage 3 process verification becomes so important.^1,2

Stage 3 is the ongoing phase of the analytical procedure lifecycle, where the laboratory verifies during routine use that the procedure remains in a state of control.^1,2 It shifts the focus away from a purely reactive model, where laboratories respond only after an out-of-specification (OOS), out-of-trend (OOT), or investigation appears, and toward a more mature way of working—one built on vigilant monitoring, traceability, and evidence-based oversight.^2,4

Why can validation alone no longer build confidence?

That shift matters because the regulatory expectation has changed. A validation package on its own is no longer enough to create confidence that a laboratory process remains fit for purpose. What matters now is whether the laboratory can show that its procedures continue to perform reliably in routine use, that sources of variability are being watched, and that decisions are supported by data rather than assumption.^1,2

As regulatory expert Peter Baker (Live Oak Quality Assurance LLC) noted in a recent webinar on Managing Method Variability: A Foundation for Risk-Based Change, “We really have to change the way we define validation.”⁴ That is a strong message for QC teams, because it shifts the conversation from validation as a completed exercise to control as an ongoing expectation. The question is no longer only whether a method was validated once. It’s now whether the laboratory can show that the method remains under control throughout its entire lifetime of use.^1,2

Vigilant monitoring: Detecting drift before it escalates

Ongoing verification matters because it gives QC teams a structured way to see developing signals earlier, understand whether procedure performance is starting to drift, and respond before variability becomes disruptive.^2,4 In a different webinar, Managing Process Performance in Stage 3 Process Validation: Regulatory Expectations for Continued Process Verification, Peter stressed the importance of laboratories understanding where variability exists within methods and procedures, and the need for monitoring programs to be implemented before those issues turn into investigations.⁴ This is the real value of Stage 3 verification. Continuous control starts with knowing what to watch, and then watching it consistently.^2,3

This broader view of monitoring is also why simply trending final results is not enough. Final results matter, of course, but on their own they do not always tell the laboratory where the problem is coming from. Peter made that point directly in the webinar when he explained that, if you are only trending the end result, “you don’t know whether the high or low result is due to the product or the method.”⁴

For QC laboratories, this is a critical distinction. When an investigation starts, time is already against you. If monitoring has been limited to final results only, the lab may not have enough visibility to determine whether the underlying signal points to the procedure, the sample, the instrument, or the process. Vigilant monitoring is what reduces that ambiguity.^2,4

Traceability: Turning data into defensible evidence

This is where traceability becomes a practical requirement, not just a compliance concept. If the laboratory can see what happened, who did what, what changed, and how results were reviewed, then data becomes easier to trust and easier to defend.^2,4 Such traceability is especially important in pharma QC, where data is used to support batch decisions, trend stability, justify investigations, and demonstrate control during inspection. A laboratory that can show its reviewers, auditors, and regulators a clear line from execution to review to conclusion is in a much stronger position than one that has to reconstruct the story after the fact. Consistent operations and compliant decision making both depend on that visibility.^2,4

This also changes how confidence is built in QC. Confidence should not come from the absence of obvious failures alone. It should come from evidence that the process is behaving as expected and that any emerging signs of drift will be seen and understood early enough to act. Peter made that point clearly in the webinar when he said that “accuracy and completeness are the two factors that FDA is really looking for.”⁴

For QC laboratories, that is the heart of the issue. Audit trails, review practices, process knowledge, trend analysis, and procedural controls matter because they support those two outcomes.^2,4 Once Stage 3 verification is viewed through that lens, its purpose becomes much clearer. It is not about creating more review for the sake of it. It is about maintaining confidence that the system remains in control and that the data can support critical product quality decisions.^2,3

From audit readiness to operational advantage

This is where Stage 3 verification moves beyond simple compliance and becomes a real operational advantage. A laboratory that monitors the right indicators is better able to detect drift early, identify repeat issues before they escalate, and distinguish isolated noise from a genuine performance signal.^2,3 That makes investigations more focused, root cause work more effective, and day-to-day operations more resilient. It also reduces the amount of firefighting that so often pulls QC teams away from productive work. Continuous control is not about adding more oversight for the sake of it; it’s about making sure the laboratory can stay in a state of control with fewer surprises and fewer decisions based on incomplete information. Continuous control means continuous confidence.^2,4

For pharma QC, that kind of confidence matters at every level: analysts need confidence that procedures are behaving as expected, supervisors need confidence that trending and review processes will highlight emerging issues in time, and quality leaders need confidence that the laboratory can explain its performance clearly and defend its decisions during inspection. That’s what a mature quality system looks like in practice. It is controlled, confident, and compliant, not because it says so, but because the evidence is there in the data, the review process, and the operational discipline that supports them.^2,4 The next step is making that evidence easier to generate, review, and defend consistently—day after day, method after method.

This is where Waters can help, enabling laboratories to generate, review, and defend that evidence through integrated systems, reproducible chromatographic technologies, traceable workflows, data intelligence tools, and expert services. The Waters approach combines purposefully designed instrumentation, scalable chemistries, informatics, and professional support to strengthen control from method design through routine monitoring.⁵

Empower Software supports audit‑ready operations by creating traceable records through audit trails and compliance‑enabled workflows, and Empower Subscriptions can make it easier to stay current with updates and related cloud‑based applications, reducing upgrade burden while maintaining compatibility and security as requirements evolve. In parallel, Waters Data Intelligence Software is designed to trend key measures, visualize risk, and support audit-readiness through data‑driven oversight.⁷

Waters Professional Services, Software Compliance Services, Instrument Qualification Services, and Empower Software‑based system services can then help laboratories deploy, qualify, validate, and maintain these approaches to support consistent operations and inspection readiness across the lifecycle.⁵

The message from regulators is clear. In modern pharma QC, confidence doesn’t come from a validation report alone. It comes from continuous evidence that the operation remains in control.^1,2 Stage 3 process verification is what helps laboratories create that evidence. It allows them to monitor what matters, reduce ambiguity, strengthen traceability, and build decisions on data rather than assumption.^2-4

That’s the foundation of consistent operations. It’s also the basis for a laboratory that is ready, every day, to show that its processes are controlled, its data is trustworthy, and its decisions can stand up to scrutiny. Built on control and backed by data.^2,4

Special thanks to Peter Baker, President, Live Oak Quality Assurance LLC, for generously sharing the regulatory insights and practical perspectives from the webinar series that informed this article.

To hear Peter Baker’s perspectives on Regulatory Expectations for Continued Process Verification in full, watch the webinar.

References

1. International Council for Harmonisation. (2023). ICH harmonised guideline Q14: Analytical procedure development. Adopted 1 November 2023.

2. United States Pharmacopeia. (2021). <1220> Analytical Procedure Life Cycle. USP–NF. Rockville, MD: United States Pharmacopeial Convention.

3. International Council for Harmonisation. (2023). ICH harmonised guideline Q9(R1): Quality risk management. Adopted 18 January 2023.

4. Baker, P. (2026). Webinar Managing Process Performance in Stage 3 Process Validation: Regulatory Expectations for Continued Process Verification.

5. Audit Ready Labs with Risk-Based Quality Control.

As clinical laboratories increasingly rely on LC‑MS for high‑throughput, multi‑analyte testing, retention time stability has become a defining factor in assay robustness, data quality, and operational confidence.

Why does retention time stability matter in clinical workflows

In research settings, small shifts in retention time may be tolerable. In clinical laboratories, they are not. Clinical LC‑MS assays are expected to deliver:

• Consistent results across long assay lifetimes
• High throughput with minimal manual intervention
• Defensible data in regulated and audited environments

When retention times drift, laboratories are forced to widen MS acquisition windows to avoid missing analytes. This creates a cascade of downstream effects, including:

• Reduced dwell time per transition
• Fewer data points across each chromatographic peak
• Increased susceptibility to interference
• Compromised quantitative precision, particularly for low‑level analytes

In large panels, even modest variability can lead to missed detections, increased manual review, and higher rates of reruns or batch failures. In practice, retention time stability is a prerequisite for scalable, high‑confidence clinical LC‑MS.

The link between retention time stability and MS performance

MModern tandem mass spectrometers are capable of extremely fast acquisition, but their performance is fundamentally constrained by chromatographic predictability.

Stable retention times enable laboratories to:

• Use narrow, well‑defined acquisition windows
• Optimize dwell time for each transition
• Increase the number of data points across peaks
• Maintain consistent peak integration and quantitation

When chromatography is unstable, mass spectrometry efficiency is lost, not because of the detector, but because the acquisition strategy must compensate for uncertainty upstream.

In this way, chromatography does not merely precede detection. It defines how effectively MS technology can be used in routine clinical workflows.

Why is UHPLC critical for retention time stability

Ultra-High Performance Liquid Chromatography (UHPLC) is often associated with sharper peaks and faster separations. In clinical LC‑MS, however, its most important contribution is reproducible chromatographic behavior over time. UHPLC systems designed for clinical use offer:

• Improved control of mass transfer and dispersion
• Greater consistency across injections and batches
• Reduced sensitivity to small changes in system conditions

These attributes translate directly into more stable retention times, which is particularly important when operating large panels in complex biological matrices. However, achieving retention time stability in a clinical environment requires more than high pressure alone. It requires systems engineered explicitly for robust, routine operation.

Retention time stability as a risk‑reduction strategy

From a clinical perspective, retention time stability is not just an analytical parameter, it is a risk‑reduction mechanism. Stable chromatography supports:

• Lower rates of assay failure and reanalysis
• Reduced manual data review
• Cleaner trending of system suitability metrics
• Greater confidence during audits and inspections

Conversely, unstable retention times can drive deviation investigations, complicate method maintenance, and erode trust in results over time.

For laboratories managing regulated workflows, chromatographic stability directly supports quality and compliance objectives.

Waters ACQUITY UPLC I‑Class PLUS IVD System: Designed for retention time confidence

The ACQUITY UPLC I‑Class PLUS IVD System has been engineered to deliver the level of retention time stability required for routine clinical diagnostics. This design philosophy reflects the realities of clinical laboratories, where consistency across shifts, operators, and time is paramount.

By providing highly reproducible retention times, the ACQUITY UPLC I‑Class PLUS IVD System enables:

• Narrow acquisition windows
• Efficient use of MS dwell time
• Reliable quantitation across high‑volume sample sets
• Sustained performance throughout the assay lifecycle

Chromatography expertise matters in clinical implementation

Retention time stability is not achieved by hardware alone. It reflects decades of chromatographic understanding applied to system design and clinical use cases.

Waters leadership in UHPLC technology, and its long experience supporting regulated laboratories has shaped how systems, like the ACQUITY UPLC I‑Class PLUS IVD System, are optimized for real‑world clinical operation.

In today’s clinical laboratory, retention time stability is not just a technical detail. It is the foundation of trust in LC‑MS results.

Discover how UPLC can elevate your clinical analysis.

That’s the real shift in the regulatory conversation. The old mindset treated validation as a staged event. You completed the studies, wrote the report, and called the method validated. That isn’t how it’s viewed anymore.

As regulatory expert Peter Baker (Live Oak Quality Assurance LLC) noted in a recent webinar series, “We really have to change the way we define validation.”¹

The expectation now is much closer to lifecycle thinking.^2,3 A method isn’t considered strong simply because it passed a validation protocol years ago. It has to continue delivering accurate, complete, and defensible data over time. That starts upstream, in development.

What are regulators really concerned about?

If you strip the language back, the concern is actually very simple. Can you trust the data? More specifically, is the data accurate, and is it complete? Those are the two questions that matter most.

Peter Baker captured that point directly when he said that “accuracy and completeness are the two factors that FDA is really looking for.”¹ Everything else, including controls, procedures, documentation, audit trails, assessments, validation studies, is there to support those two questions. If method design doesn’t support accuracy and completeness, then the rest of the package is weaker than people think.

That’s why method design matters so much. Accuracy isn’t protected just by choosing a detector or writing a sample preparation SOP. It depends on the full analytical workflow, including sample preparation, instrument conditions, data handling, analyst interaction, procedural controls, and the way variability enters the process. If those things aren’t thought through during development, then parts of method performance are being left to chance. From a regulatory standpoint, that’s a weak position to defend.

Robustness isn’t something you add at the end

This is where a lot of organizations still get into trouble. Robustness is often treated as something you confirm near the end of development, almost as a final checkbox before validation. That’s too late if the method was never built on a solid understanding of variability in the first place. If you want a method to stand up over time, you need to understand what could make it drift, what could make it fail, and what could cause it to behave differently across analysts, instruments, materials, or sites.

This work belongs in development, and if it’s not done early, it usually has to be done later—while under pressure. You start seeing signals: precision drift, unexpected system suitability failures, repeat investigations, or borderline results. That’s exactly the situation regulators don’t want to see. If variability was understood earlier, many of those downstream questions become easier to answer, and some of them never come up at all.

Better method design starts with understanding variability

So, what does better design look like? It starts with practical questions:

• Where can variability enter this procedure?
• Where are the manual steps?
• Where are the sensitive conditions?
• Where could sample preparation go wrong?
• Where could chromatographic behavior change?
• Where might hardware, software, documentation, or human interaction influence the result?

Asking these questions early and adopting a structured, risk-based development approach is vital to understanding where variability can arise, which factors are most likely to affect method performance, and how the procedure can be designed and controlled to support reliable performance across its lifecycle, in line with the principles of ICH Q14 and USP <1220>.^2-4

This means mapping the workflow, identifying the factors that can influence the reportable result, and evaluating which of those factors are most likely to matter.⁴ The method can then be designed either to control those factors or to build a scientific understanding of their impact. That’s how the discussion moves from “the method seems to work” to “we understand why it works, and where it is vulnerable”. Those are very different positions when a method has to be defended under audit scrutiny.

A trial-and-error approach isn’t enough anymore

A lot of methods are still developed with a “one factor at a time” mindset. You adjust something, see if it looks better, adjust something else, and see if it improves again. That may get you to a usable method, but it often leaves a limited understanding of how the variables interact and where the real edges of performance are. The real problem is that you can end up with a method that looks acceptable under one set of conditions but is much less well-understood than people assume.

That matters later. When a method change or adjustment is needed, the question is not just whether the change can be made. The question is whether it can be justified scientifically. If the development work didn’t generate enough knowledge, that becomes much harder. You’re left with assumptions, opinions, and incomplete rationale, and those don’t hold up well in inspections. A systematic development approach makes that justification much easier.²

The real goal is methods that are defensible by design

One of the biggest benefits of strong development is that it reduces ambiguity later. When variability appears, you are in a much better position to ask the right question: Is this coming from the method, the sample, or the process?

Without that foundation, organizations often spend time chasing the wrong root cause. They tweak the method when the real issue is upstream. They retrain analysts when the real issue is design related. They treat repeated signals as isolated events instead of seeing them as signs of a broader weakness in the method.

The takeaway is straightforward. If you want methods that stand up to audit scrutiny, don’t treat audit-readiness as something you add later. Build it into the design.^2,3 Build in the understanding of variability. Build in the scientific rationale. Build in the control strategy.^2-4 Build in enough knowledge that, when questions come later, you aren’t guessing—you are explaining.

How can Waters help you?

Waters supports a more structured, risk-based approach by combining instrumentation, scalable chemistries, integrated informatics, and expert services to help laboratories design workflows that are scientifically justified, traceable, and defensible from the start.⁵

Waters positions this approach around lifecycle-oriented analytical design, aligned with ICH Q9 and ICH Q14, so teams can make better risk-based decisions, strengthen control, and support consistent method performance over time. Professional Services, Software Compliance Services, Instrument Qualification Services, and Analytical Services can then help labs deploy, validate, qualify, transfer, and maintain those approaches in a way that supports inspection-readiness across the method lifecycle.

In the end, that’s what strong method development should deliver: not just a passed validation study, but a method that’s robust, controlled, and scientifically defensible across its lifecycle. From a regulatory perspective, that’s what long term audit readiness really looks like.

Special thanks to Peter Baker, President, Live Oak Quality Assurance LLC, for generously sharing the regulatory insights and practical perspectives from the webinar series that informed this article.

To hear Peter Baker’s perspectives on Managing Method Variability in full, watch the webinar series.

References

1. Baker, P. (2025). Webinar series Managing Method Variability: A Foundation for Risk-Based Change.

2. International Council for Harmonisation. (2023). ICH harmonised guideline Q14: Analytical procedure development. Adopted 1 November 2023.

3. United States Pharmacopeia. (2021). <1220> Analytical Procedure Life Cycle. USP–NF. Rockville, MD: United States Pharmacopeial Convention.

4. International Council for Harmonisation. (2023). ICH harmonised guideline Q9(R1): Quality risk management. Adopted 18 January 2023.

5. Waters web page: Audit Ready Labs with Risk Based Quality Control.

With capabilities like these, it’s tempting to ask a simple question: Does chromatography still matter as much as it used to?

The short answer is yes.

The longer—and more important—answer is that chromatography has never been more critical to clinical confidence.

Analytical sensitivity can’t fix a poor separation

Advances in MS have fundamentally changed what laboratories can measure. But, no matter how sensitive an MS detector becomes, it still depends on what is delivered to it. The mass spectrometer signal represents the combined effects of the analyte and co‑eluting matrix components during ionization.

Clinical samples such as plasma, serum, urine, and hair are inherently complex. Endogenous compounds, metabolites, salts, and phospholipids all compete for ionization. When chromatography fails to adequately separate target analytes from this background, the result is familiar to anyone running LC‑MS in a clinical environment — ion suppression, unstable quantitation, and results that are difficult to trust.

More analytical sensitivity does not remove these effects — it can actually amplify them. Without robust chromatography, increased analytical sensitivity simply means detecting variability more clearly.

Chromatography is not just the ‘front end’

Chromatography is often described as the front end of the LC‑MS workflow, but that framing understates its importance. In practice, chromatography defines:

• Which compounds reach the ion source together
• How consistently analytes elute over time
• How stable peak shape and retention times are across long runs
• How reproducible quantitation is from day to day

In other words, chromatography doesn’t just prepare the sample for MS detection — it establishes the conditions under which the mass spectrometer can perform reliably.

This is especially important in clinical workflows, where assays must be stable across thousands of injections, multiple analysts, and extended operating hours.

The role of UHPLC in modern clinical LC‑MS

UltraHigh Performance Liquid Chromatography (UHPLC) has become a key enabler of reliable clinical LC-MS workflows. Higher efficiency separations deliver narrower, better-defined peaks, which support both chromatographic resolution and mass spectrometry response.

Narrower peaks don’t just improve separation—they help maximize MS signal, reduce coelution, and improve signal to noise. Just as importantly, modern UHPLC systems are designed to deliver this performance consistently, not just under ideal conditions.

For clinical laboratories, that consistency matters more than theoretical peak capacity. Reproducibility, retention time stability, and robustness over long sequences directly influence assay validity and confidence in reported results.

Why clinical labs have different chromatography needs

Not all UHPLC systems are created with clinical use in mind. Research environments often prioritise flexibility and one-off performance, while clinical laboratories operate under very different pressures.

Clinical chromatography must deliver:

• Stable retention times across long sequences
• Robust injector performance over thousands of injections
• Low carryover between high and low concentrations
• Predictable behaviour across shifts, analysts, and instruments
• Minimal intervention to maintain uptime

In these settings, chromatography is not evaluated by how it performs on a good day, but by how reliably it performs every day. It can often be said that in clinical labs: the most exciting result is the one that never changes.

Better chromatography reduces operational risk

When chromatography is stable and reproducible, the benefits extend well beyond analytical performance.

Reliable separations reduce the frequency of failed runs, reanalysis, and troubleshooting. This improves throughput, protects turnaround times, and reduces stress on staff. Over time, it also supports method longevity, helping validated assays remain fit-for-purpose longer.

From a quality perspective, consistent chromatography simplifies method validation, performance trending, and audit-readiness. Variability in chromatography often becomes variability in compliance — and that is a risk that clinical laboratories cannot afford.

Confidence comes from control, not complexity

UHPLC is sometimes perceived as introducing additional complexity into clinical workflows. The opposite is true when systems are designed for regulated environments.

Well-engineered UHPLC platforms provide tighter control over solvent delivery, sample introduction, and separation efficiency. That control translates into predictability—and predictability is the foundation of analytical confidence.

When chromatography behaves consistently, laboratories spend less time compensating for variability downstream, whether that’s through data review, repeat analysis, or investigation of unexplained trends.

Chromatography is the foundation of analytical confidence

Speed and analytical sensitivity are important. But in routine clinical LC‑MS, reliability at scale is what truly determines success.

Even as mass spectrometry continues to advance, chromatography remains the factor that decides whether results are trustworthy, repeatable, and defensible. It determines how well analytical sensitivity is translated into actionable clinical data.

In modern clinical diagnostics, chromatography is not an accessory to mass spectrometry. It is the foundation on which analytical confidence is built.

Waters pioneered UHPLC and has spent more than 20 years refining it for real-world analytical demands.

See how the ACQUITY UPLC I-Class PLUS IVD System brings that legacy into today’s clinical LC-MS labs.

MaxPeak Premier Oligo OBD Columns provide a powerful solution for these workflows, now expanded to include 19mm ID formats.

These columns reduce non-specific adsorption, improve recovery, and enable consistent scalable purification from analytical to preparative workflows.

Addressing the unique challenges of oligonucleotide purification

Unlike small molecules or even peptides, oligonucleotides often exhibit:

• Strong interactions with metal surfaces due to phosphate groups
• Significant structural similarity between impurities and target sequences
• Sensitivity to method conditions that can impact recovery and purity

These factors frequently result in non-specific adsorption (NSA), poor recovery, and compromised resolution, especially during scale-up.

For oligo workflows, where isolating sufficient material for characterization or downstream use is critical, these challenges translate directly into lost productivity, wasted samples, and uncertainty in results.

Introducing columns designed for scalable oligo purification

MaxPeak Premier OBD Prep Columns extend Waters inert column technology into preparative-scale purification, enabling scientists to bridge the gap between analytical method development and purification workflows.

With the introduction of 19 mm ID columns, users can now scale to higher load capacities while maintaining performance consistency and recovery, which is critical for oligonucleotide applications that require larger material quantities.

Key capabilities include:

• Predictable scale-up from analytical (2.1 mm) to preparative formats (10 mm →19 mm)
• Improved recovery of metal-sensitive analytes such as oligonucleotides
• Reduced non-specific adsorption, minimizing sample loss and improving detection
• Efficient purification workflows without the need for column passivation  

Why do 19 mm columns matter for oligo workflows?

While 10 mm preparative columns are widely used, the addition of 19 mm ID columns expands capabilities for oligo purification, particularly when you need to:

• Scale production for preclinical or analytical characterization
• Increase load capacity for low-level or difficult-to-isolate targets
• Reduce the number of injections required for purification

This is especially important for oligonucleotides, where low recovery or repeated purification cycles can significantly impact timelines and cost.

The 19 mm platform enables:

• Higher throughput purification with fewer runs
• Consistent performance at scale
• Confidence in isolating even low-abundance oligo species

Differentiating oligos from small molecules and peptides

Although MaxPeak Premier OBD Columns support a wide range of applications, including small molecule and peptide purification, their value is particularly pronounced for oligonucleotides, which are negatively charged and more prone to metal-surface interactions/non-specific adsorption. 

MaxPeak Premier Technology greatly reduces these surface interactions, enabling cleaner separations and improved recoveries across complex oligo mixtures.

A complete OBD portfolio for flexible workflows

The MaxPeak Premier OBD Column portfolio supports a full purification workflow, including:

• Analytical-scale columns for method development
• 10 mm prep columns for initial scale-up
• 19 mm prep columns for higher-throughput purification

Standard reversed-phase chemistries (e.g., BEH C₁₈ 130 Å and BEH C₁₈ 300 Å Columns) provide flexibility to tailor separations for different oligo modalities and impurity profiles. This continuity allows scientists to scale directly from optimized analytical methods without redevelopment, reducing risk and accelerating timelines.

Enabling confident oligo purification at scale

For scientists working in oligonucleotide research, development, and manufacturing, the ability to achieve the following is essential:

• Resolve impurities effectively
• Recover sufficient material for downstream analysis
• Scale methods predictably without rework

MaxPeak Premier OBD Columns, especially with the addition of 19 mm formats, deliver on these needs by providing an inert, scalable, and high-performance purification solution.

By reducing non-specific adsorption and enabling seamless scale-up, these columns enable researchers to move from discovery to production with greater confidence, efficiency, and success.

Learn more about Waters oligonucleotide purification solutions.

Additional resources:

MaxPeak Premier OBD Columns

Purifying Problematic Compounds by RP-LC: Simplified Workflows for Better Outcomes with Inert Preparative Columns (Webinar)

Purifying Oligonucleotides: High-Efficiency Prepative Chromatography to Improve Yields and Turnaround Time (Guidebook)