Why the FDA issues Complete Response Letters: the deficiency types that recur

A Complete Response Letter (CRL) is the action FDA takes when a marketing application (NDA, BLA, or ANDA) cannot be approved in its present form. The letter enumerates every deficiency the review divisions want resolved before approval. Read across a large body of CRLs, the same problem categories surface again and again, and most letters cite deficiencies in more than one category at once. The pattern is consistent: applications rarely fail for a single dramatic reason, they fail because one or more review disciplines find the submission incomplete or unpersuasive.

Below is what recurs, ordered roughly by how frequently each category appears across CRLs, with the specific findings reviewers write most often.

The chart above counts how many distinct applications reference each substantive deficiency category. Labeling, proprietary-name, and REMS comments are omitted from the ranking because they appear in nearly every letter as near-boilerplate content (discussed separately at the end). The counts are a relative signal, not a precise tally of the single "primary" reason for each action.

1. Product quality and CMC deficiencies

Chemistry, manufacturing, and controls (CMC) problems are the single most common substantive driver of a Complete Response, and the findings are strikingly repetitive across products ^{48548449048816491492493494495496498499500503504506}. The recurring sub-themes:

• Specifications not adequate to control quality. Reviewers repeatedly say proposed release or stability specifications are too broad or unjustified, especially for critical quality attributes tied to potency or safety. Examples include a trastuzumab biosimilar where drug substance/drug product release specs for a potency-related attribute were too broad and had to be tightened ¹²⁴¹¹⁹, and a solifenacin oral suspension whose batches were not meeting the proposed specification ¹²⁷.
• Analytical methods not validated or not fit for purpose. Rapid sterility, mycoplasma, endotoxin, and identity assays for a cell therapy lacked adequate validation ¹³⁶; an omeprazole/sodium bicarbonate suspension assay could not reliably quantify bicarbonate content ¹³².
• Stability data insufficient to support the proposed shelf life. Missing long-term, in-use, freeze-thaw, or shipping data recur constantly ^118616125136.
• Container closure integrity, sterility, and microbial control gaps. FDA requires validated sterility and CCIT methods and rejects "testing into compliance" via backup sterility tests ^126136120127.
• Process validation not demonstrated. Outstanding PPQ data, in-process controls, and control-strategy justification are common ^120136128.
• Extractables and leachables assessments incomplete, including missing identification of leachables and toxicological risk assessment ^137138136.

The recurring pattern reviewers describe: a sponsor proposes a control strategy built on untested or weakly justified assays and specifications while the product has critical quality attributes tied to safety or efficacy. The standard remediation is to tighten specifications, validate methods, generate more complete stability data, and prove manufacturing consistency ^{118119123125132136137138}.

2. Facility inspection and CGMP compliance

Unresolved manufacturing-facility problems are one of the most common reasons approval is withheld, and they are often independent of the scientific merits of the application ^{48448716499505}. In a sample of these CRLs, roughly 12 of 20 cited facility inspection, CGMP, or site-compliance deficiencies as a basis for non-approval ^{124689111213141516171920}. Recurring findings:

• Objectionable conditions or CGMP deficiencies observed during inspection, which must be resolved before approval ^{6912131415161720}.
• A pre-approval inspection or re-inspection is required after issues are addressed ^{269111214151617}.
• The facility has not been confirmed in CGMP compliance, or FDA has not yet assessed whether the site can perform the listed operations ¹⁴¹⁹.
• Inspections could not be completed (for example, because of travel restrictions), delaying the action until the inspection is done ¹¹²¹⁴.
• Facility deficiencies not specific to the application but still requiring resolution at the site ²¹⁶¹⁷.
• For combination products, device quality-system compliance under 21 CFR Part 4 must be demonstrated at inspection ³⁵⁷¹⁰.

The most repeated pattern is simply that FDA observed objectionable conditions at one or more manufacturing sites and withheld approval until they were satisfactorily resolved, sometimes requiring re-inspection ^{6912131415161720}.

3. Insufficient evidence of clinical efficacy

When the deficiency is clinical, it is most often that the application does not establish substantial evidence of effectiveness ^{676971737475767879}. The recurring efficacy findings:

• Substantial evidence not demonstrated, even where FDA acknowledges unmet need or applies regulatory flexibility ^6768707374.
• A single pivotal or single-arm study is insufficient, particularly when it is not an adequate and well-controlled investigation ^67686978.
• Prespecified primary endpoints not met or not statistically significant ^71737679. In one case (MOVE-FA), the trial failed both primary and key secondary endpoints ⁷¹.
• Secondary and exploratory endpoints fail to provide convincing support ^71727679, or the observed effect is not clearly clinically meaningful even when nominally significant ⁷¹⁷²⁷⁵.
• An additional adequate and well-controlled trial is needed to establish effectiveness or isolate the drug's contribution ⁷⁴⁷⁸⁷⁹.
• Reliance on post hoc or externally controlled analyses that FDA finds unreliable, biased, or unfit for primary support ^69737778.
• Results hard to interpret because of confounding, baseline imbalance, heterogeneity, missing data, or trial-conduct problems ^{67686971727678}, and in chronic diseases, lack of durability or long-term efficacy data ⁷²⁷⁴.

4. Clinical pharmacology and dosing

A distinct and recurring bucket concerns dose justification and clinical pharmacology, separate from efficacy per se ^{478479480481482483}:

• Inadequate QT/QTc assessment. Oliceridine showed dose-dependent QTcF prolongation with limited Phase 3 ECG characterization ⁴⁷⁸; other products had no thorough QT study and an insufficient safety database to assess QT effects ⁴⁷⁹⁴⁸².
• Proposed dose or regimen not adequately supported. Oliceridine's exposure database was too small to support the proposed 40 mg maximum daily dose ⁴⁷⁸; a cytisinicline ESRD regimen was not supported by available PK and modeling ⁴⁸⁰; a testosterone undecanoate titration regimen was undermined by PK data-integrity problems ⁴⁸¹.
• Exposure-response and PK support gaps ^483481480, and bioanalytical or PK data-integrity problems that made dosing conclusions unreliable, including analytical methods that were not properly validated ⁴⁸¹⁴⁸³.

5. Bioequivalence and bioavailability (generic and bridging applications)

For generic and bridging applications, the recurring deficiencies center on failure to establish comparability to the reference product ⁴⁷⁶¹¹⁷:

• Inadequate in vitro dissolution / drug-release data to support a biowaiver or bridge, including missing n=12 profile comparisons and f2 similarity analysis across strengths ⁴⁷⁶.
• Bioequivalence not demonstrated across strengths, with some strength comparisons failing or missing entirely ¹¹⁷.
• Biowaiver requests denied because acceptable BA/BE evidence was not established ⁴⁷⁶.
• Bioavailability characterization incomplete under differing conditions such as food effect or potential CYP3A4 interactions ¹¹⁷.

6. Safety and benefit-risk

Where safety drives the action, the letters describe unresolved signals or an unfavorable overall balance ^{108109112113114116117}:

• Unresolved safety signals such as malignancy imbalances, possible drug-induced liver injury, cardiovascular/MACE risk, and thromboembolic events ^108109110.
• Unfavorable benefit-risk, with explicit language that safety concerns outweigh benefit ¹⁰⁹¹¹².
• Insufficient long-term safety data to characterize risk over the intended use ¹¹⁷.
• Immunogenicity / anti-drug-antibody concerns affecting safety or efficacy ¹¹², and safety risk from use errors or higher-than-expected systemic exposure ^113114116.

7. Nonclinical and toxicology gaps

Nonclinical deficiencies are less frequent but highly stereotyped when they appear ^{166162163164161168169}:

• Insufficient systemic-safety justification for excipients and leachables when clinical use implies systemic exposure, prompting requests for toxicological risk assessments ^162163164.
• Missing repeat-dose toxicology to cover the intended duration of use ¹⁶²¹⁶¹.
• Genetic toxicology, reproductive/developmental toxicology, and carcinogenicity gaps, sometimes a full battery for impurities or excipients ^162166164161. One letter was explicit in requiring genotoxicity, embryo-fetal, fertility, pre/postnatal, and carcinogenicity studies ¹⁶⁶.
• Route-of-administration bridging problems, where studies were conducted by a route other than the proposed clinical route ¹⁶⁴¹⁶⁶.

8. Device and human factors (combination products)

For drug-device combination products, device and human factors deficiencies recur as their own cluster ^{45845946046546847118113}:

• Human factors validation missing, incomplete, or not conducted with the to-be-marketed user interface ^{45845946046546847118113}.
• Use-related risk analyses that do not address critical use errors or the new risks introduced by device changes ^{45946046346746847047118125}.
• Device modified after HF testing with no evidence the change was revalidated ^460469470471.
• Instructions-for-Use and packaging problems, including difficult-to-open or child-resistant packaging not evaluated in HF studies ^{460462473113471}.
• Device essential-performance issues tied to safe use such as actuation force, break-loose and glide force, dose accuracy, and delivery reliability ^459463125134, and missing device stability / aging data over the product's life ^{459463469472125}.

Near-universal administrative content: labeling, proprietary name, and REMS

Almost every CRL contains labeling, proprietary-name, and REMS comments, which is why these terms match nearly the entire corpus and are not a useful discriminator of why a given application failed. FDA frequently reserves detailed labeling comment until the application is otherwise approvable, then still flags required changes ^{48548816499166501505}. The recurring administrative items:

• Prescribing information revisions, often referencing the PLR and Pregnancy and Lactation Labeling requirements and the SRPI checklist ^{83858792949899100103104}.
• Carton and container labeling revisions, including the bolded Medication Guide statement directed to the pharmacist ^{8083848687909193979899102103104}.
• Proprietary name status, typically "conditionally acceptable" pending final review, occasionally unacceptable and requiring resubmission ^{8081828485878891929394979899100101102103104}.
• REMS requirements, where a REMS is deemed necessary or the proposed REMS does not adequately mitigate the risk and must be revised, sometimes with specified elements to assure safe use, an implementation system, and assessments ^{808182838485868889909596101}.
• Safety-update requirements for resubmission, including retabulated adverse events and narratives for deaths and serious adverse events ^{485486488497166505}.

The cross-cutting takeaway

Two structural points are worth carrying into submission planning. First, product quality/CMC and manufacturing-facility readiness together account for a large share of Complete Responses, and both are frequently independent of the clinical data package: a strong efficacy story does not rescue an application if the control strategy is unproven or a site is not in CGMP compliance ^{4854846916136}. Second, when the clinical package is the problem, the recurring failure is not a subtle statistical miss but a fundamental one: a single inadequate or not-well-controlled study, an unmet primary endpoint, or reliance on external controls and post hoc analyses that cannot support substantial evidence of effectiveness ⁶⁷⁶⁹⁷⁸. For combination products and 505(b)(2) filings, device human factors and bridging/certification adequacy add further recurring failure modes that sit outside the traditional clinical review ^458459485501.