Combination Product Labeling Under 21 CFR Part 3: Auditing the Drug–Device Gray Zone With AI

Autoinjectors. Drug-eluting stents. Photosensitizer-activation systems. Prefilled syringes with proprietary needle safety mechanisms. The combination products category touches more of the pharmaceutical and medical device supply chain than most compliance teams realize — and the global market is tracking toward an estimated $200 billion by 2030.

Yet most QA organizations are built around a single lane. Drug teams audit to 21 CFR Part 201. Device teams audit to 21 CFR Part 801. And at the seam between those two frameworks — where a single labeling document has to satisfy both — deficiencies concentrate in ways that neither team is structurally positioned to catch.

FDA’s Office of Combination Products (OCP), established under MDUFMA 2002 and operating under Section 503(g) of the FD&C Act, has consistently identified labeling as one of the highest-frequency deficiency areas in combination product submissions. The errors are almost never about ignorance of one regulation. They’re about the intersection of two regulations that no single reviewer was tasked to manage simultaneously.

AI-augmented audit tools are beginning to change that calculus. But before getting to the technology, let’s be precise about what 21 CFR Part 3 actually establishes — because in my experience working with regulated manufacturers across the US, most teams are genuinely fuzzy on the framework, and that fuzziness is expensive.

What 21 CFR Part 3 Actually Establishes (And Where Teams Get It Wrong)

21 CFR Part 3 governs how FDA assigns combination products to the appropriate lead center — CDER for drug-led products, CDRH for device-led products, or CBER for biologic-led products — based on the product’s primary mode of action (PMOA). When center assignment is uncertain, manufacturers can file a voluntary Request for Designation (RFD) with the OCP. FDA has 60 days to respond under 21 CFR 3.7(f), and that determination becomes the regulatory foundation everything else is built on.

The PMOA assignment is consequential for labeling because it establishes which center’s regulations are primary — but not exclusive. A combination product with a drug PMOA puts CDER in the lead, making 21 CFR Part 201 the governing drug labeling framework. But 21 CFR Part 801 doesn’t disappear. The device constituent still has to satisfy device labeling requirements. CDER simply takes responsibility for ensuring those requirements are also met during review — which is not the same as saying the manufacturer is relieved of the obligation to satisfy them.

Here’s the trap: most QA teams interpret “CDER-led” as “drug labeling review only.” The device labeling requirements become an afterthought, something to address if the reviewers happen to raise it. And FDA’s reviewers, who are center specialists themselves, sometimes don’t catch the device labeling gaps because they’re also organized around single-center expertise.

The result is a labeling package that survives pre-market review but accumulates deficiencies that surface during post-market inspections or, worse, through a competitor complaint routed to OCP.

The Four Labeling Zones Where Deficiencies Concentrate

From a practical audit standpoint, combination product labeling failures cluster in four zones. Each one is addressable with systematic review — but only if you’re holding both regulatory frameworks in working memory at the same time.

Zone 1: Directions for Use

21 CFR 201.57 and 21 CFR 801.109 both address directions for use, but with different scope and specificity requirements. Drug labeling under 21 CFR 201.57 requires structured Prescribing Information for prescription products; device labeling under 21 CFR 801.109 requires adequate directions for use for the device constituent. When the same document — typically an Instructions for Use (IFU) or a combined PI/IFU — must serve both purposes, reviewers frequently satisfy one standard while inadvertently creating deficiency against the other. Language that meets the completeness criteria of 21 CFR 201.57 may fail the proximity-to-device-use standard of 21 CFR 801.109, and vice versa. It’s a conflict, not a gap — and conflicts are harder to catch than missing sections.

Zone 2: Safety and Warning Language

Contraindications under 21 CFR 201.57(c)(5) and hazard warnings under 21 CFR 801.4 have overlapping but legally distinct purposes. Drug contraindications are risk-based clinical determinations; device hazard warnings are physical safety statements about the device constituent itself. A prefilled syringe label can be meticulous on drug contraindications while remaining silent on the needle-stick risk profile of the delivery mechanism — which is a device labeling deficiency under 21 CFR 801.4 whether or not anyone in the drug regulatory group noticed. These gaps are entirely routine in combination product warning letters.

Zone 3: Identity, Nomenclature, and Intended Use

Combination products create naming complexity that neither a drug team nor a device team typically owns well alone. Generic name requirements under 21 CFR 201.10 apply to the drug constituent; device nomenclature governed by FDA classification panels applies to the device constituent. When a single product trade name appears on a label, it has to work for both. And when marketing proposes product family names that blur the drug/device boundary — or when a device name implies a drug indication — you’re one complaint or one agency review away from an unintended intended-use reclassification. This is the highest-consequence zone and the one most often left to marketing teams without regulatory compliance consulting input.

Zone 4: Manufacturing Source and Component Disclosures

Where the drug constituent ends and the device constituent begins must be explicit in labeling for certain combination product types, particularly prefilled container systems and kit configurations. FDA guidance documents issued between 2017 and 2022 reinforced expectations that labeling clearly delineate which GMP system governed which constituent part. Labeling that obscures this boundary — or that’s silent on component-level manufacturing attribution — routinely appears in OCP enforcement correspondence.

Why Traditional Regulatory Compliance Consulting Misses This

I want to be direct about something: the billing model that governs most regulatory compliance consulting engagements is structurally misaligned with combination product labeling audit. A client hires a drug regulatory consultant for the CDER pathway and a device regulatory consultant for the CDRH pathway. Neither of those consultants is being compensated to audit the labeling document from the other center’s perspective. Each delivers expert work within their lane. The combined labeling package passes sequential review and fails holistic review.

This isn’t a critique of any individual consultant. It’s a structural problem baked into how regulatory work is organized. Combination product labeling audit requires simultaneous, not sequential, multi-framework review — someone or something that holds 21 CFR Part 201 and 21 CFR Part 801 in working memory at the same time and actively looks for points of conflict rather than confirming individual compliance.

AI-augmented audit tools have a genuine structural advantage here. Not because AI knows the regulations more deeply than a senior regulatory affairs professional — experienced humans still win that comparison on nuance and strategy. But because a well-configured AI review system doesn’t carry center-specific bias. It checks both regulatory frameworks against the same document passage in a single pass, flags tension points where satisfying one requirement creates deficiency against another, and produces a prioritized gap report with citation to specific CFR sections and applicable FDA guidance documents. That’s not something a sequential human review workflow can replicate economically.

At Aurora TIC, our DeepGMP document review module was built to handle exactly this kind of multi-framework labeling audit. It reads labeling documents against concurrent regulatory frameworks, identifies conflicts (not just gaps), and presents findings organized by regulatory priority and remediation complexity. It isn’t a substitute for expert judgment on regulatory strategy decisions — those require human experience with the OCP process and center-specific precedents. But it eliminates the structural blind spot that comes from center-siloed review.

Building an AI-Augmented Combination Product Labeling Audit Protocol

The following five-step protocol reflects how we structure combination product labeling audits for manufacturer clients. Traditional execution by a human regulatory team runs 6–10 weeks for a complex labeling package. With AI-augmented review handling steps 2 and 3, most teams compress that to 10–14 days without reducing audit depth.

1. Confirm your PMOA determination is documented. Before auditing a single labeling document, verify that the PMOA determination is either anchored in an RFD response from OCP (within the 60-day clock under 21 CFR 3.7(f)) or documented internally with explicit legal and scientific rationale under 21 CFR 3.7. An undocumented PMOA means your labeling framework has no regulatory foundation. Fix this first.

2. Build a dual-framework requirement matrix. Map every applicable section of 21 CFR Part 201 and every applicable section of 21 CFR Part 801 to the document structure. Each requirement gets a row; the document location addressing it gets a column; a gap/conflict flag gets a third column. AI tools can auto-generate this matrix from an uploaded labeling document in minutes — a task that takes a human reviewer 2–3 days manually.

3. Audit for conflicts, not just gaps. Run both regulatory frameworks against the same document passages simultaneously. A missing section is a gap and is obvious. A passage that satisfies 21 CFR 201.57 while creating deficiency against 21 CFR 801.109 is a conflict and is subtle. AI-powered review catches the conflicts that sequential human review structurally misses.

4. Run a marketing language overlay. Pull every marketing-influenced text element — product name, indication summary, taglines, patient-facing descriptions — and test each against FDA’s intended use doctrine. A single phrase that implies a drug indication on what FDA has classified as a device constituent can trigger reclassification without anyone intending it. This step is non-negotiable for any combination product with both direct-to-professional and direct-to-patient labeling components.

5. Human expert review of AI-flagged items. The AI audit produces a prioritized finding list. A regulatory affairs professional with combination product experience reviews each finding, makes regulatory strategy decisions on gray-zone items, and closes the loop. AI handles the mechanical multi-framework cross-reference; human expertise handles the strategic judgment. The ratio of AI to human time should be roughly 70/30 — and that ratio is what makes the 10–14 day timeline achievable.

The One Question That Reveals Your Audit Gap

Combination product labeling is one of the few areas in FDA regulatory affairs where deep expertise in one framework actively creates a blind spot for another. When your team completes a combination product labeling review, ask one question: Was there a single moment in that process where someone was looking at a 21 CFR Part 201 requirement and a 21 CFR Part 801 requirement at the same time, on the same document passage, specifically looking for conflicts rather than individual compliance?

If the answer is no — and for most organizations it is — you have a structural gap that no amount of sequential expert review can close. AI-augmented audit tools don’t replace the expertise your team already has. They change the architecture of how that expertise gets applied, so the seam between two regulatory frameworks gets the same scrutiny as the centers on either side of it.

Written by Sam Sammane, Founder & CEO, Aurora TIC | Founder, Qalitex Group. Learn more about our team

Reserve early access to our AI audit tools — including DeepGMP for multi-framework labeling review. Contact us

Related from our network

• Raw Material COA Review and Supplier Qualification — Qalitex Laboratories provides ISO 17025-accredited testing and document review for pharmaceutical and supplement manufacturers navigating FDA supplier qualification requirements.
• Canadian GMP Compliance and Health Canada NHP Testing — Androxa supports Canadian manufacturers with GMP audits, Health Canada NHP submissions, and the cross-border regulatory compliance documentation that US-exported combination products often require.