How FDA Assesses Anti-Drug Antibody Immunogenicity in BLA Reviews

How FDA assesses anti-drug antibody immunogenicity in BLA reviews

FDA evaluates immunogenicity in a biologics license application (BLA) along two lines. First, reviewers decide whether the anti-drug antibody (ADA) and neutralizing antibody (NAb) assays can reliably detect the antibodies that matter. Second, they decide whether the antibodies detected change exposure, efficacy or safety. The findings then determine what the label says and whether FDA issues postmarketing requirements (PMRs) or commitments (PMCs). This article follows those steps using FDA guidance, BLA review documents and approved labels.

The guidance framework

Three FDA guidances define reviewer expectations:

Guidance	Status and date	Role in the review
Immunogenicity Assessment for Therapeutic Protein Products	Final, August 2014	Risk-based framework, risk factors, clinical consequences, sampling ¹⁰⁷¹⁰⁸
Immunogenicity Testing of Therapeutic Protein Products: Developing and Validating Assays for Anti-Drug Antibody Detection	Final, January 2019	Tiered assay design, validation parameters, clinical data analysis ¹⁷³¹²⁴
Immunogenicity Information in Human Prescription Therapeutic Protein and Select Drug Product Labeling: Content and Format	Draft, February 2022 (not for implementation)	Where and how immunogenicity appears in labeling ²⁰³²¹²

A risk-based, case-by-case approach

The 2014 guidance recommends a case-by-case, risk-based approach to evaluating and mitigating immune responses. The assessment should start early in development, be revisited as the program progresses, and reflect both the likelihood and the potential severity of a response for the specific product and population ¹⁰⁷¹⁰⁸.

Patient-related factors. These include immune competence, immune suppression or activation, age, prior exposure or allergy, route, dose and dosing frequency. For replacement therapies they also include genetic differences and tolerance to the endogenous protein. A move to a new population calls for a new assessment ¹⁰⁹¹¹⁰¹¹¹.
Product-related factors. These include molecular structure and sequence, impurities, glycosylation or PEGylation, formulation, container interactions, and storage or handling. Changes to these factors may warrant new immunogenicity testing ¹¹²¹¹³¹¹⁴¹¹⁵¹¹⁶.

Clinical consequences FDA looks for

Efficacy and PK. Neutralizing antibodies can block activity. Both neutralizing and non-neutralizing antibodies can change clearance, half-life and efficacy ¹¹⁷.
Hypersensitivity and anaphylaxis. FDA recommends identifying anaphylaxis by clinical criteria regardless of the presumed mechanism, and cautions that ADA status alone does not predict a reaction ¹¹⁸¹¹⁹.
Cross-reactivity with endogenous proteins. Antibodies can cross-react with and neutralize an endogenous counterpart, which could remove a critical physiological function ¹²⁰.

Assay validation expectations

Multi-tiered testing

The 2019 guidance describes a tiered approach. Clinical samples are first screened with a sensitive assay designed to detect both low- and high-affinity ADA. Screen-positive samples then go to a confirmatory assay for drug-specific binding. Confirmed positives are characterized further with titration and neutralization assays, plus other characterization when warranted ¹⁷⁴. Titer measures the magnitude of the response. Neutralization assays assess whether ADA interfere with the product's interaction with its target ¹⁷⁵. Titer, persistence and neutralizing activity may tell a reviewer more than ADA incidence alone ¹²⁵.

Key validation parameters

Parameter	FDA recommendation (2019 final guidance)
Screening cut point	Determined statistically from roughly 50 treatment-naive samples from the study population, each tested by at least two analysts on at least three days. One approach uses a one-sided 90% lower confidence bound for the 95th percentile of negative-control responses, which assures a false-positive rate of at least 5% ¹⁷⁶. A floating cut point is common. A fixed cut point is discouraged because negative-control means can vary over a study ¹⁷⁷.
Sensitivity	Sensitive enough to detect ADA before they reach levels associated with altered PK, PD, safety or efficacy. FDA recommends at least 100 ng/mL for screening and confirmatory IgG/IgM assays, subject to risk and prior knowledge. Neutralization assays may not reach that level. IgE assays should reach the high pg/mL to low ng/mL range ¹⁷⁸. Sensitivity should be reported per mL of undiluted matrix, accounting for the minimum required dilution, and assessed with the expected on-board drug present ¹⁷⁹.
Drug tolerance	Assessed early by spiking positive-control antibody into ADA-negative samples with and without varying drug concentrations. Acid dissociation may improve tolerance but does not suit every case. Collecting samples at trough can reduce drug interference ¹⁸⁰.
Specificity	The assay should detect ADA and not product, soluble target, nonspecific endogenous antibodies or assay reagents. Rheumatoid factor interference should be addressed where relevant ¹⁸¹.
Positive controls	A purified product-specific positive control (affinity-purified polyclonal or monoclonal). Control antibodies may not represent patient ADA and can overestimate sensitivity ¹⁸²¹⁸³. A low-positive system-suitability control should test positive consistently in both the screening and confirmatory tiers ¹⁸⁴.
Matrix	Sensitivity tested in the same matrix at the relevant dilution and, where applicable, with the same anticoagulant as clinical samples ¹⁸⁵.

Sampling design

Pretreatment samples should be collected from all subjects. A sample about 7 to 14 days after first exposure can show an early IgM response, 3 to 6 weeks is generally best for IgG, and IgA may peak at about 2 to 3 weeks ¹⁸⁶. With repeated dosing, sampling should continue at appropriate intervals through the trial and about 30 days after last exposure, or about five half-lives after for long-half-life products. Where serious ADA consequences are a high risk, sampling should continue until ADA return to baseline ¹⁸⁷. The 2014 guidance adds that samples should be drawn before dosing, that sampling should be more frequent early in chronic use, that unscheduled samples should be taken after suspected immune-related events, and that serum banking is advisable ¹²¹¹²². Drug levels should be measured in the same samples to identify interference with ADA detection ¹⁸⁸¹²³.

How reviewers judge assay adequacy in practice

BLA reviews show that assay adequacy often decides whether immunogenicity can be interpreted at all.

Assays found suitable, with limits documented. For HADLIMA (adalimumab-bwwd), the assessor found the ADA and NAb assays suitable for their intended purpose. In the rheumatoid arthritis study, the ADA assay had 63 ng/mL sensitivity and drug tolerance of 20 µg/mL for SB5 and 40 µg/mL for EU-Humira, with a confirmatory cut point of at least 44% inhibition. The NAb assay had drug tolerance below 0.5 µg/mL. The assessor noted it still detected NAbs in more than 50% of rheumatoid arthritis subjects ²⁴⁰²⁴¹.
Assay found unsuitable. For EBANGA (ansuvimab-zykl), the reviewer found validation inadequate because of a suboptimal cut point, missing drug-tolerance and selectivity data, and poor specificity. FDA did not issue a PMC to revalidate the assay. It told the sponsor that any future clinical immunogenicity assessment or labeling revision would need fully validated screening, confirmatory and titer assays, shown fit for Ebola-positive samples with a cut point confirmed in that population ²⁴².
Cut point too high, incidence underestimated. For ELELYSO (taliglucerase alfa), deficiencies raised in a Complete Response Letter included a positive control too far from the cut point, a confirmatory cut point that needed to be re-established, drug-tolerance testing that did not cover low antibody concentrations, and the need to estimate IgE assay sensitivity ²⁵¹. A separate review found that an unreasonably high confirmatory cut point underestimated incidence. As a result, effects on PK and efficacy were not adequately determined, NAb activity was inconclusive and safety impact was not determined ²³⁷.
Drug interference. For LEQEMBI (lecanemab-irmb), the accelerated-approval review reported ADA in 63/154 patients (40.9%), with NAbs in 16/63 (25.4%) of ADA-positive patients. Serum drug could interfere with both assays, and their drug tolerance was below most relevant predose concentrations, so incidence may have been underestimated. The reviewers found insufficient information to characterize effects on PK, PD, safety or effectiveness and called for improved, validated assays in the confirmatory study ²⁴³²⁴⁴.
Stricter in-study cut points. For RETACRIT (epoetin alfa-epbx), reviewers concluded that the PK/PD similarity conclusions were unaffected by updated, more stringent in-study assay cut points ²⁴⁶.

Clinical impact on PK, efficacy and safety

The 2019 guidance calls for analyses that relate ADA status and titer to PK, PD, efficacy and adverse event data. These analyses should take into account drug levels in tested samples and the effects of pre-existing, treatment-boosted and treatment-induced antibodies ¹²⁴. Review documents show the range of conclusions FDA reaches:

Product	ADA / NAb findings	FDA's assessment of impact
FASENRA (benralizumab)	Treatment-emergent ADA in 13% to 15% on drug vs 4% on placebo; most ADA-positive patients had NAbs	Higher clearance and higher blood eosinophils in ADA-positive patients, but no observed effect on efficacy or safety ²⁴⁷
PIASKY (crovalimab-akkz)	Treatment-emergent ADA in 60/191 (31.4%) treatment-naive and 43/184 (23.4%) switch subjects	Clearance 25% higher in subjects with medium/high titers. About 3% (11/375) lost pharmacological activity with reduced exposure, with variable clinical impact that included loss of hemolysis control. NAb results not reviewed because of assay issues ²⁴⁸
EMRELIS (telisotuzumab vedotin-tllv)	Treatment-emergent ADA in 60/269 (22%); NAbs in 23/60 (38%) of ADA-positive patients	ADA/NAbs associated with a 17% decrease in clearance. FDA found insufficient data to determine effects on safety or effectiveness. The applicant had separately argued there was no apparent impact ²⁴⁹
RETACRIT (epoetin alfa-epbx)	IV comparative study: 0.4% treatment-induced ADA and 0.0% neutralizing ADA in both arms (301 and 304 participants); no NAbs detected across the reported studies	No apparent ADA impact on PK, PD, efficacy or safety ²⁴⁵²⁴⁶

Two patterns appear in these reviews. FDA separates a measurable PK effect from a clinically meaningful effect. FDA also states plainly when the data cannot support a conclusion, and does not default to "no impact." The EMRELIS review shows the difference between FDA's assessment and the applicant's position ²⁴⁹.

Endogenous cross-reactivity

When a product shares sequence with an endogenous protein, FDA may require targeted testing. For ADLYXIN (lixisenatide), the review called for testing ADA-positive samples to determine NAb incidence and whether antibodies cross-react with and neutralize endogenous GLP-1 or glucagon. The assays had to be formally validated and reviewed by FDA before sample analysis, and the study reports had to evaluate effects on safety, PK, PD and efficacy ²⁵⁰.

Postmarketing requirements and commitments

When assays are not adequate at approval, FDA commonly asks for improved assays and retesting of banked samples after approval:

Product	Requirement or commitment	Rationale given
Hemlibra (emicizumab)	CMC PMCs to develop and validate an ADA assay that detects antibodies at the emicizumab levels expected during sampling, and an assay for neutralizing capacity ⁶⁵⁶⁶	The proposed drug-tolerance target was low compared with the reported trough concentration ⁶⁷⁶⁸
Leqembi (lecanemab)	PMRs under section 505(o) to improve ADA assay sensitivity to at least 100 ng/mL at the expected trough drug level, improve NAb assay sensitivity and drug tolerance, and then use the improved assays to assess PK, PD, safety and efficacy ⁶⁹⁷⁰	FDA identified an unexpected serious immunogenicity risk that spontaneous reporting and the available postmarket system would not adequately assess ⁷¹
Trutakna (atacicept)	PMC to develop a sensitive, drug-tolerant NAb assay, then analyze banked samples and assess impact on PK and efficacy ⁷²	The binding ADA assays were adequately validated, but no NAb assay was available ⁷³
Zynlonta (loncastuximab tesirine-lpyl)	PMC 4029-6: develop a drug-tolerant neutralizing ADA assay, test all confirmed ADA-positive banked samples and correlate results with safety and efficacy ⁷⁴⁷⁵	To characterize the clinical relevance of neutralizing antibodies ⁷⁵
Crysvita (burosumab)	Proposed PMR to reanalyze banked serum with a more drug-tolerant validated ADA assay, characterize neutralization and assess effects on PK, efficacy and safety ⁷⁶⁷⁷	To obtain ADA data with the improved assay ⁷⁶⁷⁷
Ultomiris (ravulizumab-cwvz), supplemental BLA	PMCs 4082-1 and 4082-2: improve drug tolerance of the screening and confirmatory ADA assay, then reanalyze samples from Study ALXN1210-PNH-304 ⁷⁸⁷⁹⁸⁰	To determine ADA incidence with an assay that tolerates the drug levels expected at sampling ⁷⁹⁸⁰

PIASKY's review also notes a planned PMC to optimize and validate the NAb assay ²⁴⁸.

Immunogenicity labeling

FDA's proposed approach (2022 draft guidance)

Location. For biological products with immunogenicity data, the study summary goes in 12.6 Immunogenicity under Clinical Pharmacology, with cross-references where clinically relevant information appears elsewhere ¹⁹⁰²¹³. Subsection 12.6 is meant to be the consistent location for ADA incidence and PK/PD effects. Other sections are reserved for clinically significant effects ⁹³.
Standard statement. When the methodology is adequate, the subsection opens by stating that observed ADA incidence depends heavily on assay sensitivity and specificity, and that differences in assay methods prevent meaningful comparison with other studies, including studies of the same or related products. ADA incidence and NAb incidence follow, with both positive and negative findings summarized ¹⁹⁶.
Inadequate methodology. If the evaluation method cannot establish ADA incidence, FDA proposes a statement that the information is insufficient to characterize the ADA response and its effects on PK, PD, safety or effectiveness ²¹⁴²¹⁵.
Clinically significant effects elsewhere in the label. ADA-associated adverse reactions go in Adverse Reactions (clinical trials or postmarketing experience) ¹⁹¹. Clinically significant changes in effectiveness, including efficacy results by ADA status, go in Clinical Studies ²¹⁶. Clinically significant ADA-associated risks are summarized briefly in Warnings and Precautions ¹⁹⁴. When an effect is unknown, that is generally stated in 12.6 ²¹⁷¹⁸⁹¹⁹³.
No cross-product comparisons. The 2019 assay guidance warns that comparing incidence across products can mislead. A direct comparison requires a head-to-head study using an assay shown to have equivalent sensitivity and specificity for both products ²¹⁸.

What current labels look like

Placement varies across approved labels. ADUHELM uses 6.2 Immunogenicity, and VYVGART uses 6.3. HEMLIBRA, RYSTIGGO, LEQEMBI and TECENTRIQ HYBREZA use 12.6 ²²²³²⁴²⁵²⁶²⁷. Labels commonly carry the assay-dependence caution. TECENTRIQ HYBREZA extends it to other atezolizumab and hyaluronidase products ²⁵²⁶.

Product (section)	Incidence	Stated clinical impact
HEMLIBRA (12.6)	ADA 5.1% (34/668); in vitro NAbs 2.7% (18/668)	Decreased plasma concentrations in 0.6%. One patient with NAbs and decreased concentrations lost efficacy with breakthrough bleeding ³⁰³¹³²
RYSTIGGO (12.6)	ADA 37% (49/133); NAbs 21% (28/133)	Up to 60% lower trough concentrations in ADA-positive patients. No apparent clinically meaningful efficacy impact, and similar safety profiles ²⁹
TECENTRIQ HYBREZA (12.6)	ADA 20% (43/221); NAbs in 54% (21/39) of ADA-positive patients	Clearance 29% higher, not expected to be clinically significant. Effect on effectiveness unknown. No clinically significant safety effect in the first six months ³⁴
LEQEMBI (12.6)	18-month IV study: ADA 3.4% (30/883); NAbs 1.9% (17/885)	Lower mean trough concentrations. No clear clinically meaningful impact on PD, safety or efficacy, though low incidence limits a conclusive assessment ³³
ELAHERE (12.6)	Treatment-emergent ADA 7% (28/423); NAbs 6% (24/423)	Effect on PK, efficacy or safety unknown because antibodies were infrequent ³⁵
ADUHELM (6.2)	Up to 0.6% (15/2,689)	NAbs not quantified. Data too limited for definitive conclusions ²²²⁸

Consistent with the label caution, these figures are not a cross-product ranking of immunogenicity ²⁵²⁶.

Biosimilars and interchangeable products

For 351(k) applications, immunogenicity is assessed as part of the comparison with the reference product.

FDA describes a stepwise, risk-based approach in which clinical immunogenicity testing sits alongside analytical, PK/PD and safety data. Even when PK and PD similarity may be enough to address clinically meaningful differences, a full safety and immunogenicity evaluation should still be done ¹⁴⁸¹⁴⁹.
In interchangeability switching studies, immunogenicity is assessed descriptively. This covers ADA and NAb incidence and titers, their effects on PK, PD, safety and efficacy, and assay sensitivity in the presence of drug. The same assay should be used in the switching and non-switching arms ¹⁵⁰¹⁵¹¹⁵²¹⁵³.
For insulin, draft guidance says a comparative clinical immunogenicity study generally would not be needed if a state-of-the-art comparative analytical assessment leaves little or no residual uncertainty about immunogenicity. A study may still be needed case by case ¹⁵⁴¹³²¹⁴⁶.
An October 2025 draft guidance says comparative analytical data, an appropriately designed PK similarity study and an immunogenicity assessment together may be enough without a comparative efficacy study. It does not say the immunogenicity assessment can be omitted ¹⁵⁵¹⁴¹.
For labeling, the 2023 draft biosimilar labeling guidance generally recommends incorporating relevant reference-product immunogenicity information with appropriate changes. When that information is included, the assay-comparison caution should come before the data ¹⁹⁷.

Practical takeaways for sponsors

Assay drug tolerance is a common point of failure. Several of the PMRs and PMCs above require ADA or NAb assays that work at the drug levels expected during sampling ⁶⁵⁶⁹⁷²⁷⁴⁷⁶⁷⁹.
Validate the NAb assay before the pivotal readout. Missing or inadequate NAb assays led to unreviewed NAb data or postmarketing work for PIASKY, Trutakna and Zynlonta ²⁴⁸⁷²⁷⁴.
Bank samples. Many postmarketing obligations depend on reanalyzing stored trial samples with improved assays ⁷²⁷⁴⁷⁶⁸⁰.
Expect FDA to separate PK effects from clinical effects. Reviews and labels routinely report clearance or trough changes and then make a separate judgment on efficacy and safety, sometimes stating that the impact is unknown ²⁴⁷²⁴⁹³⁴³⁵.
Assay adequacy shapes the label. If FDA judges incidence unreliable, the proposed labeling language says the data are insufficient to characterize the ADA response ²¹⁴²¹⁵.