For regulatory and CMC teams managing NDA or BLA submissions for injectable drugs and delivery devices, understanding where CDER concentrates its information requests and Complete Response deficiencies is essential for submission readiness and lifecycle planning. Recurring review themes signal both agency expectations and gaps that applicants have consistently failed to address prospectively.
This analysis synthesizes CDER review activity from 2024–2025 across injectable presentations—including vials, prefilled syringes, autoinjectors, pens, and wearable injectors—identifying the drug-quality and device essential-performance topics that generated the most frequent deficiencies, with representative approved or non-approved products cited to illustrate each theme.
Want Rhizome's help on your own question? Try it for free.
Recurring CDER review themes for injectable drug-device combination products, 2024-2025
Across 2024-2025 CDER reviews of injectables and their delivery systems, information requests (IRs) and Complete Response deficiencies clustered into a predictable set of drug-quality (CMC) and device essential-performance topics. The pattern is consistent whether the applicant filed an NDA or a BLA, and whether the presentation was a vial, prefilled syringe (PFS), autoinjector, pen, or wearable injector. Below are the themes that recur most often, each with representative products rather than an exhaustive IR list.
Drug quality (CMC) themes
1. Container closure integrity: prove sensitivity to sub-20-micron breaches
The single most repeated CMC request was to demonstrate that the container closure integrity test (CCIT) is sensitive enough to detect the small breaches that could permit microbial ingress, generally framed as detection of defects smaller than 20 microns. FDA raised this for ANKTIVA (nogapendekin alfa inbakicept-pmln), noting the method in 3.2.P.5.2 had not been shown to have sufficient sensitivity and asking for routine positive controls 210. The same demand recurs almost verbatim across biosimilar and originator injectables: SIMLANDI (adalimumab-ryvk) 212, NYPOZI (filgrastim-txid) 214, JUBBONTI/WYOST (denosumab-bbdz) 218, HERCESSI (trastuzumab-strf) 220, EPYSQLI (eculizumab-aagh) 222, and MERILOG (insulin aspart-szjj), where the stability CCIT was judged to lack sensitivity and needed to detect breaches below 20 µm 232.
A closely related sub-theme is inadequate validation of dye-ingress CCIT methods. For VYLOY (zolbetuximab-clzb), FDA said the non-compendial dye-ingress method was validated only for specificity and detection limit and still needed precision, repeatability, intermediate precision, reproducibility, and robustness data 228. For STEQEYMA (ustekinumab-stba), the dye-ingress method was not validated for reproducibility between different analysts, and FDA asked for multi-analyst evidence or a spectrophotometric alternative with updated validation 413. FDA also frequently coupled these requests with an instruction to substitute CCIT for sterility testing on annual stability samples 212214218220, and for ALHEMO (concizumab-mtci) additionally asked for evidence of primary-container integrity after assembly into the pen injector 216.
2. Sterility assurance, bacterial endotoxin, and low endotoxin recovery
Aseptic-processing and sterility-assurance IRs were near-universal for sterile injectables, typically bundling requests for processing/hold times, bioburden and endotoxin sampling limits, sterilizing-filter retention validation, equipment sterilization/depyrogenation, media fills, shipping validation, CCI, and qualification of bioburden/sterility/endotoxin methods. This full package was requested for SIMLANDI 233, PYZCHIVA (ustekinumab-ttwe) 224, AHZANTIVE (aflibercept-mrbb) 265, and NYPOZI 266. A recurring, more specific request was for low endotoxin recovery (LER) studies, because certain formulations can mask endotoxin detectability in the USP <85> BET; FDA asked for spiking/recovery-over-time data in undiluted product across SIMLANDI, PYZCHIVA, AHZANTIVE, NYPOZI, EPYSQLI, and HERCESSI 233224265266235276.
MERILOG illustrates how these coalesce into hard deficiencies: FDA cited unclear endotoxin-removal steps, inadequate validation of the microbial-contamination in-process control, incomplete descriptions of the microbial and endotoxin release tests, missing CCI data for crimping pressure, insufficient CCIT sensitivity, and missing LER data 236. STEQEYMA drew sterility-assurance deficiencies on multiple fronts: post-use integrity-test acceptance criteria based on an insufficient basis, stopper sterilization validation lacking the parameters needed to demonstrate a 10^-6 sterility assurance level, worst-case-location gaps in glass-vial depyrogenation/sterilization studies, and requests for additional media-fill data 268229223217. OTULFI (ustekinumab-aauz) had residual concerns requiring additional real-world shipping data, sterilizing-filter pressure control, capping/crimping validation, endotoxin-method optimization, and CCIT revalidation 272.
3. Extractables and leachables: incomplete profiles and insufficient shelf-life coverage
E&L IRs centered on whether the extractables profile was complete enough to design the leachables program, and whether leachables were followed across shelf life. The clearest deficiency was for the ZURNAI autoinjector (nalmefene hydrochloride): FDA found it implausible that only three non-volatile and no volatile/semi-volatile extractables were detected, judged the study design likely incapable of producing a complete profile, and asked the sponsor to re-run extractables in 5% IPA, 50% IPA, and pH 3 and pH 9 buffers with 12-24 h reflux, ground/smaller samples, and LOQs at or below the analytical evaluation threshold, then repeat leachables accordingly 181193. FDA also said the proposed leachables timepoints were too sparse to establish a shelf-life trend and required at least three commercial batches at ICH Q1A timepoints under accelerated and long-term conditions using validated methods 193.
The shelf-life-duration angle recurs for TYENNE (tocilizumab-aazg), where FDA said the leachables study needed to be extended through 60 months and identified a manufacturing material as a potential extractables/leachables source warranting additional toxicological assessment 189192203. DAWNZERA (donidalorsen sodium) and BKEMV (eculizumab-aeeb) show the mirror-image outcome, where FDA scrutinized the AET/PDE approach and compounds detected above threshold but ultimately concluded there were no leachables of toxicological concern 191190196186194.
4. Particulate matter and compendial specification gaps
Particulate-matter comments split between routine parenteral labeling statements (inspect for visible particulates before use, seen across many products) and genuine CMC specification gaps. For TRYPTYR (acoltremon, NDA 217370), FDA asked the applicant to bring the drug-product specification into line with ophthalmic compendial standards, recommending USP <790> for visible particulates and reference to USP <771>, after which the applicant added visible inspection and subvisible testing 8770. For LYMPHIR (denileukin diftitox-cxdl), visible/appearance and subvisible particulate matter were flagged as product- or process-related concerns tied to the manufacturing process, formulation, and container-closure system 71. On the device-facing side, YESAFILI (aflibercept-jbvf) drew a particulate/ocular-safety deficiency: the sponsor had not performed intravitreal injection irritation testing with particulates per USP <789> and had not stated conformance with ISO 10993-1, with FDA also requesting support for intraocular injection specifications covering USP <789> particulates, endotoxins, sterilant residuals, and ocular irritation 128.
5. Specification acceptance criteria and stability/shelf-life justification
FDA repeatedly asked sponsors to justify drug-substance and drug-product stability specifications, particularly where stability acceptance criteria differed from release criteria under the same analytical method, and to consider all batches and all timepoints rather than only release and end-of-shelf-life data. This request pattern appears for ALHEMO 151175, WINREVAIR (sotatercept-csrk) 154, and IMULDOSA (ustekinumab-srlf) 167. STEQEYMA drew related deficiencies: an unjustified clarity release/stability specification, an instruction that stress/accelerated data should not be used to set long-term acceptance criteria, and a shelf-life limit for acidic peaks that full-term stability did not support 155165170. ZIIHERA (zanidatamab-hrii) carried a postmarketing requirement to update release and stability specifications with a validated C1q-binding ELISA and justified acceptance criteria as a CDC surrogate 174. YORVIPATH (palopegteriparatide) is the crossover case where content/assay acceptance limits intersected the device: FDA warned that the revised assay and delivered-volume limits could allow dose variability and dose overlap, and that no stability data existed on delivered dose 161.
Delivery-device essential-performance and ISO themes
6. Dose accuracy, delivered volume, and ISO 11608 conformance
The flagship device-quality theme was delivered-dose/volume accuracy measured against ISO 11608. For BYNFEZIA PEN (octreotide acetate, NDA 213224), CDRH found the proposed dose-volume accuracy specification too wide and non-compliant with ISO 11608-1:2022, and asked the sponsor to tighten it to meet both clinical acceptance and the standard 237. YORVIPATH is the most serious example: FDA judged the pen's dialing resolution unsuitable for the small dose volumes, called the delivered-dose/volume specifications too permissive, warned of dose overlap between dose levels and between pens from different lots, and noted the sponsor lacked design-verification data showing the revised delivered-dose specification could be met at release and stability, with about half of developmental batches failing under the proposed criteria 241251161255257. Other 2024-2025 combination products (ZURNAI 239, QFITLIA (fitusiran sodium) 242, ANDEMBRY (garadacimab-gxii) 240244248) show ISO 11608-1/-3/-5 cited as design inputs with essential-performance requirements found acceptable, illustrating the expected baseline against which the deficiency cases stand out.
7. Break-loose/glide force, activation force, and release-testing of force EPRs
FDA frequently pressed sponsors on whether force-related essential performance requirements (EPRs) were controlled by release specifications and supported by stability data. For IDACIO (adalimumab-aacf), FDA said it was unclear what specifications the sponsor intended for PFS break-loose/glide force and safety-system activation/locking, and asked why these could not be implemented immediately as release acceptance criteria; the sponsor then adopted the design specifications as acceptance criteria 108112. For ANDEMBRY, FDA cited a deficiency because EPR stability data (including break-loose and glide/injection force) had not been provided on three device lots; the sponsor supplied aged three-lot data and FDA marked it resolved 114116. For the ZURNAI autoinjector, FDA commented that the activation force appeared high and suggested narrowing the specification so intended users could activate it easily 118. XROMI (hydroxyurea) shows FDA deferring the acceptability of break-loose, extrusion (glide), and withdrawal-force specifications to OPQ 109111, while BYNFEZIA flagged possible out-of-spec glide force and injection force not reported in the submission 4.
8. Injection/delivery time and time-between-clicks: accelerated-aging and stability failures
A distinct device-stability theme was delivery-time performance drifting out of specification on aging. For HYMPAVZI (marstacimab-hncq), FDA issued a mid-cycle deficiency because 6-month accelerated-aging data failed acceptance criteria for delivery time and delivered volume, and later observed long injection times on some samples, requesting root-cause analysis to prevent recurrence in commercial product; the verification set covered cap-removal torque, needle-guard/activation force, needle extension, delivery time, and time-between-clicks 339334332230. QFITLIA showed long injection times of 18 to 27 seconds in a small fraction of samples at 24 months, prompting FDA to request continued root-cause analysis 329334. These cases signal that FDA expects delivery-time EPRs to hold across the full shelf life, not just at release.
9. Design controls, EPR identification, and traceability (21 CFR 820.30 / 21 CFR Part 4)
Where the device constituent's design-control package was thin, FDA asked for the complete 21 CFR 820.30 framework: design inputs/outputs, verification, validation, risk management file, a traceability matrix, and explicit identification of EPRs. SOFDRA (sofpironium bromide) drew a full request including a device-constituent design-requirements document, verification traced to design inputs under intended-use conditions, design validation, a traceability matrix, and specific metered-pump EPRs (pump delivery/dose weight, gel content uniformity, actuation force) 3130. RISVAN (risperidone ISM) is a stark example: FDA said the sponsor had not provided the device-constituent description or a summary of its design-control system and requested the full design history file, purchasing controls, and CAPA information 42. JUBBONTI/WYOST 3241 and YORVIPATH 35 received parallel design-control and control-strategy requests, including that stability and shipping documentation demonstrate EPRs are maintained at expiry. TRYNGOLZA (olezarsen sodium autoinjector) drew a design-verification deficiency for excluding audible/visual feedback from the EPRs without rationale, plus a gap in risk-evaluation and traceability detail 54.
10. Human factors validation, use-related risk, and critical-task classification
Human factors IRs most often turned on task criticality and IFU clarity rather than a failed validation study. VYALEV (foscarbidopa; foslevodopa) generated the largest set: FDA repeatedly reclassified as critical several tasks the applicant had labeled non-critical (warnings/knowledge tasks, troubleshooting, hand hygiene, adhesive application, changing rate versus administering an extra dose), tying misclassification to potential overdose, underdose, or infection, and recommending training or IFU revisions where use errors occurred 295298299304296312313317314. Methodology concerns appear for ANDEMBRY, where the moderator directed participants to the IFU in a step not identified in the submitted protocol, though FDA accepted it because the tasks were also observed in simulated use 294. Device-interface use errors recur for TYENNE (autoinjector orientation/activation) 297, EBGLYSS (lebrikizumab-lbkz) (holding the pen at the site until completion) 307, YORVIPATH (fully pressing the injection button) 308, and DAWNZERA (unexpected needle-cap removal force and twist-versus-pull confusion) 327328. RINVOQ LQ (upadacitinib) shows FDA issuing IRs for the use-related risk analysis and clarification of why a bottle adapter was excluded from the user interface 318.
11. EPRs in the post-approval stability protocol, shipping, and 95/95 sampling
A cross-cutting device theme was FDA's insistence that device EPRs be monitored in the post-approval stability program and demonstrated after shipping, with statistically justified sample sizes. For ALHEMO, FDA asked that dose accuracy, activation force, hold force, and injection time be added to drug-product release specifications and included in the post-approval stability protocol 449. For both YORVIPATH and HYMPAVZI, FDA asked for the number of samples tested at each stability timepoint, sized to give 95% confidence and 95% reliability that the device performs through shelf life, and for YORVIPATH specifically that stability testing verify performance of the assembled cartridge-plus-pen at expiry 218.
What the pattern implies for sponsors
Two through-lines dominate. On the drug-quality side, FDA is asking sponsors to prove the sensitivity and validation rigor of their sterility-adjacent methods (CCIT sub-20-µm detection, dye-ingress precision/reproducibility, LER) and to justify every specification against full-term stability across all batches. On the device side, the recurring gap is treating essential performance requirements (dose accuracy against ISO 11608, force parameters, delivery time) as a one-time verification exercise rather than as release specifications supported by aged, statistically powered stability and shipping data, backed by a complete 21 CFR 820.30 design-control and traceability file. Human factors friction concentrates on task-criticality judgments and IFU clarity. The biosimilar cohort (multiple ustekinumab, adalimumab, aflibercept, denosumab, and filgrastim products) shows these requests issued in near-identical language, which makes them highly predictable and largely preventable before submission.