Changing the delivery device on an approved subcutaneous combination product: what FDA has expected, and when it accepted less

When a marketed subcutaneous drug-device or biologic-device combination product changes its delivery presentation, whether that is a move from a prefilled syringe (PFS) to an autoinjector or pen, a new needle configuration, or a redesigned firing mechanism, FDA does not apply one fixed evidence package. Reviews in Drugs@FDA and CBER show the Agency runs a gap analysis against the approved product and then asks for evidence in whichever of four domains the change actually touches: human factors and use-related risk, pharmacokinetic (PK) or delivery comparability, clinical safety and local tolerability, and CMC / essential drug delivery output (EDDO) performance ²⁶²⁸. The size of the package tracks the size of the difference in the user interface and the fluid path. This article maps the recurring patterns and the specific precedents where FDA required a full study versus where it accepted a comparative analysis or effectively waived testing.

The framework FDA applies to a device change

For a change such as PFS to autoinjector, FDA's own analysis identifies the differences that drive the data ask: a new secondary container closure, a different injection method, and a different user interface ²⁶. Each difference maps to a question:

• A new user interface raises whether the design still supports safe and effective use, a human factors / use-related risk question ²⁶²⁸.
• A different injection method can change the PK / delivery profile, injection consistency, and variability ²⁶.
• A different rate of delivery to the target tissue can change the local adverse reaction profile, a clinical safety question ²⁶.
• Assembly into an autoinjector can affect the drug and container, so CMC / product quality (degradation, sterility, syringe break resistance, functionality across shelf life) is in scope ²⁶.

FDA states explicitly that the data needed to support a combination product design change "may involve more than an HFE assessment," so sponsors should not assume human factors alone will close the file ²⁸. The exact package is set by the identified gaps ²⁶²⁸.

The threshold analysis gate: what decides whether new human factors testing is needed

Before demanding a new validation study, FDA expects the sponsor to run threshold analyses that characterize how different the new interface actually is from the reference. There are three ^18111250:

1. Labeling comparison: a side-by-side, line-by-line comparison of the prescribing information, Instructions for Use (IFU), and device constituent-part descriptions ¹⁸¹²⁵⁰.
2. Comparative task analysis: a systematic comparison of the manual and cognitive steps users perform with each product, focused on where use error could arise ¹⁸¹²⁵⁰.
3. Physical comparison of the device constituent parts ¹¹²⁵⁰.

These analyses identify the differences and where the same or similar risks apply ¹⁸⁵. If meaningful differences appear, FDA evaluates the associated use-error risk and may then ask for a comparative use human factors study, designed to confirm that use-error rates for critical tasks affected by the differing external critical design attributes are not worse than the reference product's ^252182187.

Where the change introduces no new or impacted critical tasks per the use-related risk analysis (URRA), the submission can fall into a lower human factors category with no validation testing ¹⁸⁵. For a modified device that does introduce or impact critical tasks, the sponsor must address whether the existing risk controls remain acceptable; if they do, a written rationale can substitute for new validation data, and if they do not, a new (often scope-limited) validation study is expected ^18818328. In practice FDA scopes any new study to the affected use scenarios and critical tasks rather than re-running the entire program ²⁸. FDA also expects risk to be mitigated through device design first, not labeling or training alone ²³⁷.

A key nuance for critical-task identification: tasks are critical if they could affect dosing (over/under/missed dose), affect administration (wrong site, improper preparation), or result in harm ³. For time-urgent, emergency-use injectors, most or all tasks may be critical because of the life-saving context, which raises the bar for that device class ³.

Human factors and use-related risk: the recurring pattern

Across the reviews, human factors validation plus a URRA is the one domain FDA almost always engages on, but the depth varies with the interface change.

• Full validation required and accepted. For Praluent (alirocumab), FDA required a comprehensive URRA and an HF validation protocol using representative users, realistic training, and final-design devices, plus a differentiation session to confirm users could tell the pen/autoinjector apart from other pens and from the other strength ^9297103109. Observed errors (incomplete injections, failure to activate) were judged manageable through IFU and labeling revisions rather than device redesign, and FDA accepted the results after those revisions ^9410495. Auvi-Q (epinephrine autoinjector) similarly ran three simulated-use studies that CDRH found systematically evaluated use-related risks and validated user performance of the highest-priority tasks ²¹⁵²²².
• Supplemental study required for an untrained-user gap. For Tremfya (guselkumab), when the presentation changed to the One-Press patient-controlled injector, FDA accepted the sponsor's HF studies but, because the first study used only trained participants, recommended a supplemental HF study with untrained participants before concluding the mitigations and IFU were adequate ²⁰¹.
• Validation deemed not needed / effectively waived. For Fasenra (benralizumab), FDA concurred that an HF validation study was not required because the accessorized PFS was identical to a standard PFS, the updated URRA identified no new or unique risk, and use was healthcare-professional-only in a familiar setting; use was considered validated through completed clinical studies and satisfaction of user requirements ¹⁵⁹¹⁶⁰. For Emgality (galcanezumab), DMEPA concluded HF validation studies were not needed for the proposed PFS and autoinjector ¹⁶⁹. For the Humira Pen firing-button change (Supplement 381), FDA did not request HF testing at all because the change was not visible to the patient and did not alter the user interface; the file rested on device performance testing showing force-to-fire and performance within specification ^299301303.
• Editorial labeling changes without new validation. For Aimovig (erenumab), FDA found some IFU wording clarifications were needed but explicitly characterized them as editorial and implementable without additional HF validation studies ⁵⁸⁷⁰. FDA reached the same conclusion for a naloxone nasal-spray second-dose instruction change, stating it could be implemented without submitting additional HF validation testing for Agency review ³²¹.

PK and delivery comparability: when a bridging study is expected

Whether FDA wants a PK bridge depends on whether the injection method or fluid path could change exposure.

• PK / bioequivalence bridge required and accepted. This is the most common bridge for a PFS-to-autoinjector switch when the delivery mechanism changes. Dupixent's pen/autoinjector supplement was supported by a relative-bioavailability comparison to the PFS that fell within the 80% to 125% no-effect boundary ¹⁵⁰. Aimovig relied on a relative bioavailability / bioequivalence study (Study 20140477) that FDA's clinical pharmacology review said supported bridging to the marketed PFS and autoinjector ⁶⁸, with a population-PK approach accepted to compare injection sites ⁷⁴. Emgality's autoinjector was supported by a bioequivalence study (CGAQ) between PFS and pen ¹⁶⁸¹⁷². Nucala (mepolizumab) provided a PK comparability study (204958) showing comparable systemic exposure for the autoinjector and safety syringe versus the approved lyophilized product ³³⁴⁷. Auvi-Q ran a comparative-bioavailability study (INT0802) versus EpiPen and confirmed equivalent exposure ²²⁰²³⁰.
• PK bridge not required; comparability accepted on device similarity. For Auvi-Q, FDA also stated that demonstrating bioequivalence was not strictly required because the needle length, gauge, and injection force were already similar to EpiPen, so drug delivery was expected to be similar, even though the sponsor ran the study anyway ²²⁰. A biowaiver was granted for the lower 0.15 mg strength, supported by the 0.3 mg BE study plus proportional similarity and needle-dimension similarity ²³⁰²²⁷.
• No PK bridge sought. Where the primary container closure and drug pathway are unchanged and only the outer mechanism changes, several reviews contain no PK bridging requirement at all, for example the Humira firing-button change ²⁹⁹³⁰⁰, Fasenra's accessorized PFS ¹⁵⁹, and Ozempic's pen use-error supplement ³⁴¹³⁴³. FDA's general position is that a PK bridge may be unnecessary where the change raises no device-comparability concern, though the Agency does not reduce this to a single codified rule ²¹¹²¹².

Clinical bridging: usually not required for the device change alone

A separate clinical efficacy bridge for a device change is the exception, not the rule.

• Not required. Dupixent, Aimovig, Emgality, Nucala, and Auvi-Q were all cleared for their device changes without a dedicated clinical efficacy bridge; FDA relied on the known efficacy of the drug plus PK and human factors evidence, and for Auvi-Q FDA stated plainly that no separate clinical efficacy trials were conducted ^{2311546816746}.
• Required. Repatha (evolocumab) is the counter-example. At end-of-Phase 2, FDA agreed with simulated-use HFE, home-use studies, and PK/PD comparability between PFS and autoinjector, but said the proposed open-label extension substudies would not be sufficient to show whether home use affected clinical effectiveness, and required controlled Phase 3 data with LDL-C as an outcome ⁸⁸⁸⁹. Praluent's reviewers likewise flagged that if both autoinjectors and prefilled syringes were to be marketed, adequate Phase 3 clinical data would be expected, though the device-change review itself centered on human factors ⁹³⁹⁹.

When FDA accepted a comparative analysis or waived a study: the precedent table

Product	Device change	HF study	PK bridge	Clinical bridge	What FDA accepted / waived
Fasenra	Accessorized PFS (HCP use)	Not required 159160	None 159	None 159	HF validation waived; identical standard PFS, no new URRA risk 160
Emgality	PFS + autoinjector pen	Not required (DMEPA) 169	BE study CGAQ 168172	None 167176	Comparability accepted; same primary container, device used in pivotal trial 167176
Humira Pen (S-381)	Firing-button redesign	Not requested 299301	None 299	None 299	Comparability accepted on performance testing; change not visible to user 299303
Dupixent	PFS to pen/autoinjector	Validated 148151	Rel. BA within 80-125% 150	None (actual-use comparison) 154	Comparability accepted; same drug, formulation, volume, container 150
Aimovig	PFS to autoinjector	Validated; IFU edits only 7158	BE study supports bridge 68	None	PK bridge is the main bridge; IFU edits without new HF 5870
Nucala	Lyophilized to liquid AI / safety syringe	Validated + actual-use 323646	PK comparability 3347	None	Package accepted; commercial device representative 3537
Ozempic	Pen use-error supplement	Justification + prior validation 341343	None	None	Reliance on Saxenda pen validation + URRA accepted 341342
Praluent	PFS + pen/autoinjector	Full validation required 9297	Not for interface change	Phase 3 if both marketed 9399	Accepted after HF testing + labeling revision 94104
Tremfya	PFS to One-Press injector	Supplemental HF (untrained users) 201	None in record	None in record	Leveraged prior HF data; required added untrained-user study 193201
Repatha	PFS to autoinjector / on-body	Simulated-use HFE 89	PK/PD comparability 89	Controlled Phase 3 required 88	Comparability alone NOT accepted; clinical bridge demanded 8889
Auvi-Q	Novel epinephrine autoinjector	Simulated-use validated 215222	Comparative BA vs EpiPen 220230	None 231	BE "not required" on needle similarity; 0.15 mg biowaiver 220230

Emergency-use and time-critical devices raise the bar

For rescue products, the calculus shifts. Because most or all tasks can be critical in a time-urgent context, FDA leans harder on human factors validation and is less willing to treat partial-dose or activation errors as acceptable ³. Auvi-Q's approval rested on a full simulated-use HF program plus PK comparability, with FDA acknowledging the sham-device limitation but still finding the program adequate to identify and mitigate design risk ²¹⁵²²². Naloxone reviews show FDA requiring both a PK bridge and human factors data, not one or the other: Kloxxado (8 mg nasal spray) was supported by a two-study PK bridge requiring comparable or higher early-phase exposure, yet still drew a complete response citing lack of human factors data ^325328334. By contrast, Praluent reviewers noted the product was not a rescue medication, which supported treating some partial-dose risk as acceptable in the benefit-risk context ^95104105.

Needle configuration and the EDDO reassessment

A needle change is handled through the essential drug delivery output framework. FDA treats needle length as an EDDO when the length is validated to reach a specific tissue independent of user technique ¹³⁶, while injection depth is treated as user-dependent and generally not a device-controlled EDDO ¹³⁴. When a design changes, the sponsor must re-identify and reassess the relevant EDDOs, which for injectors and prefilled syringes can include deliverable volume, cap removal force, injection forces (breakloose/glide force and needle-safety activation force), injection time, dose accuracy, and needle length ^111134136125. Verification is expected after preconditioning that simulates shipping and storage stressors, and validation under actual or simulated use with representative production units, extending to shelf life where applicable ^112123130133. Wegovy's file illustrates the granularity: FDA reviewed a needle-cover override force against ISO 11608-5 and accepted a pressure-pain threshold rationale, and accepted device-performance bracketing and bridging across a comparable semaglutide formulation based on similar density and viscosity, while still requiring irritation and sensitization testing on the final finished product ^315312311.

Practical takeaways for the submission

• Lead with the gap analysis and the three threshold analyses. They are the document FDA uses to decide whether anything beyond a rationale is needed, and a clean labeling / task / physical comparison is often what supports a "no new validation" outcome ^181250183.
• Match evidence to the interface change. A change invisible to the user (Humira button) tends to close on performance testing; a new injection mechanism (Dupixent, Aimovig, Emgality, Nucala) tends to need a PK bridge plus HF validation; a genuinely novel or home-use shift (Repatha) can pull in controlled clinical data ^2991506888.
• Keep the primary container and fluid path constant where possible. Same-container arguments repeatedly carried the PK question and, combined with a benign URRA, supported HF waivers (Emgality, Fasenra) ¹⁷⁶¹⁶⁰.
• Do not assume a single argument clears the file. FDA states the package "may involve more than an HFE assessment," and naloxone shows a strong PK bridge does not excuse missing human factors data ²⁸³²⁵.
• Scope, do not repeat. Where a new study is warranted, FDA scopes it to the affected critical tasks and use scenarios rather than a full re-validation, and will accept leverage of prior validation when the users and devices are representative ²⁸¹⁹³.