Antibody Titers, Immunobridging and Correlates of Protection in FDA Approvals

Antibody titers, immunobridging and correlates of protection in FDA approvals

FDA has approved many products on the basis of immune-response endpoints rather than clinical disease outcomes. Almost all of them are vaccines reviewed by CBER. The approvals fall into four distinct evidentiary categories, and FDA's justification differs for each:

1. Established immune markers that support traditional approval, such as anti-HBs ≥10 mIU/mL for hepatitis B vaccines and opsonophagocytic activity (OPA) for invasive pneumococcal disease.
2. Immunobridging, where effectiveness is inferred from noninferior immune responses compared with a population or product in which clinical efficacy was already shown. FDA has used this for COVID-19 vaccines, RSV vaccines, pneumococcal conjugate vaccines and smallpox vaccines.
3. Immune responses "reasonably likely to predict clinical benefit", which support accelerated approval with a required confirmatory study. Examples include influenza HI titers, chikungunya neutralizing titers and OPA for pneumococcal pneumonia.
4. Animal Rule immunobridging, where human antibody levels are linked to survival in animal challenge models. The anthrax vaccines are the main examples.

The CDER record is much thinner. CDER-reviewed anthrax antitoxin monoclonal antibodies were approved under the Animal Rule using cross-species drug exposure bridging. That is not a vaccine-style immune-response endpoint. In the CDER labels and reviews examined, serum neutralizing titers or immunobridging were not identified as a basis for any approval.

Regulatory framework: validated surrogates vs. "reasonably likely" surrogates

FDA guidance separates two levels of certainty. A validated surrogate endpoint can support ordinary (traditional) approval. A surrogate that is reasonably likely to predict clinical benefit is a lower level of certainty and supports accelerated approval only, with studies required to verify clinical benefit ^233236237. For vaccines, FDA describes clinical disease-endpoint efficacy trials as the most direct way to show effectiveness. It also notes that identifying an immune correlate of protection in such a trial can help interpret later immunogenicity-bridging studies ²³⁴²³⁵.

FDA treats correlates as specific to the product and the evidence. The COVID-19 vaccine guidance says a potential surrogate depends on the vaccine's characteristics, and that a marker established for one vaccine platform cannot simply be assumed to apply to another ²³⁹. The same guidance says understanding of immune responses that predict COVID-19 protection was limited and evolving. New vaccine programs were therefore told to pursue traditional approval based on direct efficacy evidence ⁴¹¹. Accelerated approval based on an immune-response surrogate "may" be considered when there is enough evidence that the surrogate is reasonably likely to predict benefit. FDA said such a marker should ideally come from human efficacy studies of clinical disease ⁴¹².

For influenza, FDA guidance places HI antibody responses in the "reasonably likely" category. They may be acceptable for accelerated approval even though prospective effectiveness studies had not identified a specific HI titer associated with protection, and confirmatory postmarketing effectiveness studies are required ²³⁸. For pandemic influenza vaccines, FDA suggests these success criteria for adults under 65 and for children: a lower bound of the two-sided 95% CI of at least 40% for seroconversion and at least 70% for an HI titer ≥1:40 ⁴¹⁵. Seroconversion is defined as a rise from <1:10 to ≥1:40, or at least a fourfold rise when the baseline titer is ≥1:10 ⁴¹⁸. FDA calls these current recommended targets and says they are not the only acceptable ones, because understanding of immune responses to pandemic strains is still evolving ⁴¹⁶.

Category 1: established immune markers supporting traditional approval

Pneumococcal conjugate vaccines: OPA and IgG as bridging markers

FDA accepted OPA as an established immunologic parameter for inferring protection against invasive pneumococcal disease (IPD) through comparison with licensed vaccines. For PREVNAR 20 in adults, this supported traditional approval for IPD caused by all 20 serotypes. The 13 shared serotypes were compared with PCV13 and the seven new serotypes with PPSV23 ⁶¹⁰⁶¹¹. Adults 18 to 49 and 50 to 59 were then bridged to PREVNAR 20 recipients aged 60 to 64. For each serotype, the lower bound of the 95% CI for the OPA GMT ratio had to exceed 0.5 ¹⁵⁰¹⁵¹. For VAXNEUVANCE, the statistical review describes licensure as inferred effectiveness through immunobridging to PREVNAR 13. It cites PREVNAR 13's later confirmation of efficacy in the CAPiTA study and the infeasibility of new efficacy studies when uptake of approved vaccines is high ¹⁵³.

In infants, FDA used the percentage reaching a serotype-specific IgG concentration (classically ≥0.35 µg/mL) after dose 3 as the endpoint. FDA was explicit that this is a bridging threshold, not an established protective level for every serotype. The PCV15 review says it followed a WHO recommendation to bridge to PCV7 efficacy, and the Prevnar 13 memo states that no protective IgG level had been established for any serotype ⁴⁹⁹⁵⁰⁰. The noninferiority margin was a −10 percentage-point lower bound ^501504505. When serotypes missed the criterion, FDA looked at the totality of the immune data rather than treating the miss as definitive:

• Prevnar 13: serotypes 6B, 9V and 3 missed, and FDA also considered functional OPA responses ⁵⁰²⁵⁰³.
• PREVNAR 20: six serotypes missed, and IgG GMC and OPA data supported effectiveness for them ⁵⁰⁷.

Category 2: immunobridging to demonstrated efficacy

COVID-19 vaccines

FDA presented immunobridging to VRBPAC as a comparison with a regimen whose clinical efficacy had already been shown. It was not presented as proof that a particular titer guarantees protection. FDA stated that neutralizing antibodies were clinically relevant but that no specific titer had been established to predict protection ³³⁷³³⁸. For pediatric programs, FDA proposed noninferiority margins of −10% for the seroresponse rate (SRR) difference and 1.5-fold for the GMT ratio. FDA cautioned that a lower pediatric dose might require better understanding of the relationship between titer and protection, and encouraged direct efficacy evaluation where feasible ³³⁹³⁴⁰.

• SPIKEVAX, adolescents 12 to 17: Day 57 neutralizing ID50 GMT and SRR were compared with adults 18 to 25 from the pivotal efficacy trial. Success required a GMT ratio lower bound >0.67 and an SRR difference lower bound >−10 percentage points, and both were met ¹³⁷¹⁴¹. Adolescent COVID-19 incidence was a secondary, descriptive objective ¹⁴².
• COMIRNATY, adolescents 12 to 15: Noninferiority of the neutralizing GMT ratio compared with 16- to 25-year-olds required a lower bound >0.67 ¹⁴⁴¹⁴⁵. The reviewer identified immunobridging, not the exploratory efficacy objective, as the basis for inferring effectiveness ¹⁴⁶.
• NUVAXOVID, adolescents and boosters: The adolescent GMT ratio needed a lower bound >0.67 and a point estimate >0.82, and the seroconversion-rate difference needed a lower bound >−10 percentage points. All three were met ¹⁴⁹. FDA noted the adolescent study was not powered to establish efficacy ¹⁴⁸. Booster effectiveness was inferred from noninferior responses after the booster compared with after the primary series ¹⁴⁷.
• Booster and pediatric EUAs: FDA's briefing documents show that it applied these criteria consistently. The Moderna 50-µg booster met the GMT ratio criterion but did not meet the prespecified SRR criterion ^342343344345. FDA characterized the Janssen booster analysis as a post-hoc analysis of 17 participants, not prespecified hypothesis testing ³⁴⁶. For BNT162b2 in children 5 to 11, and in children 6 months to 4 years, effectiveness was inferred by bridging to 16- to 25-year-olds ^347348349.

RSV vaccines in adults under 60

RSV vaccine indications for adults at increased risk of lower respiratory tract disease were extended to younger ages through supplements: AREXVY to ages 50 to 59 and 18 to 49, and ABRYSVO and MRESVIA to ages 18 to 59 ^{431432433434435436437}. For these extensions, the bridge was to older adults in whom efficacy had been shown:

• AREXVY: At-risk adults 50 to 59 were compared with recipients aged 60 and older on RSV-A and RSV-B neutralizing GMTs and SRRs. Noninferiority required an upper CI bound ≤1.50 for the GMT ratio (60+ divided by 50 to 59) and ≤10 percentage points for the SRR difference. Both subtypes met these criteria ^{665666668669670}.
• ABRYSVO: At-risk adults 18 to 59 (Study 4) were compared with adults 60 and older (Study 3). Adjusted GMT ratios were 1.57 for RSV A and 1.52 for RSV B, and SRR differences were 5.1 and 8.3 percentage points. All lower bounds cleared the 0.667 and −10 point margins ^672674676677.
• MRESVIA: The available documents describe the planned P303 immunobridging design but do not give the success criteria or results ⁷⁰².

Smallpox and mpox vaccines

The JYNNEOS approval shows FDA requiring immunobridging plus animal efficacy evidence when there is no correlate. The endpoint was noninferiority of vaccinia-neutralizing PRNT GMT to ACAM2000, with a 0.5 margin ²⁹⁵. FDA considered animal protection data critical because the two vaccines differ and no neutralizing-antibody level predictive of smallpox protection had been established ²⁹⁵²⁹⁷. ACAM2000's original approval used noninferiority to Dryvax for vaccine take and Day 30 neutralizing GMT ^300301302. Its later mpox indication rested on macaque challenge protection plus human observational data, because direct human efficacy testing was infeasible ^298303304.

Category 3: accelerated approval on immune responses reasonably likely to predict benefit

The reviews give several types of justification:

• Influenza: shortage and seriousness. In an early HI-based review (AFLURIA), FDA stated that HI ≥1:40 was not an established immunologic correlate of protection ⁷³. It nonetheless treated the co-primary HI endpoints as reasonably likely to predict benefit, citing the risk of serious influenza illness during an anticipated vaccine shortage, with postmarketing clinical-endpoint studies required ²⁴⁷⁴. For FLUCELVAX in ages 4 to <9, traditional approval was not supported because noninferiority to Fluvirin failed for A/H3N2. The observed response was still accepted as reasonably likely to predict benefit ³⁸.
• Influenza: correlate-of-risk analysis. For MFLUSIVA, adults 50 to 64 received traditional approval based on a relative vaccine efficacy of 26.6% (95% CI 16.7 to 35.4) compared with a standard-dose vaccine ⁶²⁹⁶³⁰. Adults 65 and older received accelerated approval. The P304 efficacy data were only supportive in that group because the standard-dose comparator was not preferentially recommended for older adults ⁶³². FDA cited the biological plausibility of HAI and a prespecified correlate-of-risk analysis that linked Day 29 HAI titers to illness for A/H1N1 and A/H3N2. The analysis was inconclusive for B/Victoria because of few cases ³⁵.
• Pneumococcal pneumonia: feasibility and ethics. FDA said immunogenicity alone was insufficient for a noninvasive-disease indication because antibody levels had not been shown to predict prevention of pneumonia. It also found that an active-controlled efficacy trial was impracticable and that a U.S. placebo-controlled trial would be unethical ⁶⁰⁷⁶⁰⁸. For CAPVAXIVE, FDA noted that OPA measures antibody-mediated bacterial killing and reflects a plausible mechanism of protection, while no accepted protective threshold exists in adults ⁷⁸.
• Chikungunya: thresholds from nonhuman primates. Both chikungunya thresholds were derived from passive transfer of human immune serum in nonhuman primate challenge models, with prevention of viremia as the outcome ⁷⁵⁴⁹. FDA explicitly recorded the limits of this evidence. Prevention of disease had not been shown in the model, and other nonclinical studies found protection against viremia without protection in joints ^7547657658. In the VIMKUNYA phase 3 studies, 97.8% of vaccinees aged 12 to <65 and 87.3% of those ≥65 reached the threshold, compared with about 1% of placebo recipients. Both studies met a ≥70% lower-bound criterion ^654655656.

Risk with surrogate-based licensure: IXCHIQ

IXCHIQ shows what can happen when a surrogate-based approval is followed by unresolved safety signals. FDA's suspension memorandum records several steps:

• Approval: Accelerated approval on November 9, 2023, on a surrogate endpoint. No confirmatory study had yet demonstrated vaccine effect ⁵⁹².
• Pause: FDA and CDC recommended pausing use in people 60 and older in May 2025 ⁵⁷⁹⁵⁸⁰.
• Label change: Labeling was revised on August 6, 2025, to narrow the indication and add serious chikungunya-like illness warnings ⁵⁸¹⁵⁸².
• New findings: By August 15, 2025, there were 32 serious adverse-event cases, including 21 hospitalizations and three deaths ⁵⁸⁴⁵⁸⁵.
• Suspension: The license was suspended effective August 22, 2025 ⁵⁸⁷⁵⁸⁸. FDA cited the absence of completed confirmatory trials and of directly demonstrated effectiveness ⁵⁸³⁵⁹³. CBER said it would propose withdrawal of the accelerated approval unless Valneva timely requested that the matter be held in abeyance ⁵⁹¹.

Category 4: Animal Rule immunobridging

FDA's Animal Rule guidance asks sponsors to choose immune markers and to characterize the animal immune response associated with disease prevention. Those data then guide a human dose and schedule that produce analogous responses ³¹². The guidance warns that antibody levels alone are not necessarily a sufficient correlate, and that some pathogens may need cellular-response markers as well ³¹³³¹⁴.

• BioThrax (anthrax PEP): The endpoint was the proportion with toxin-neutralizing antibody (TNA) NF50 ≥0.56 at Day 63. That threshold corresponded to a 70% probability of survival in rabbits ¹¹⁰¹¹¹. FDA considered TNA a functional measure and concluded that animal thresholds predicting 70% survival gave a reasonable basis for predicting human benefit ¹¹².
• CYFENDUS (AV7909): There were two co-primary endpoints: NF50 ≥0.56 at Day 64, and noninferiority to BioThrax on the proportion reaching NF50 ≥0.29. The ≥0.29 threshold was accepted based on at least 70% protection in nonhuman primates ^11511686. FDA noted that the TNA assay is species-independent, so it allows direct comparison between animals and humans ¹¹⁷. Both criteria were met ¹¹⁸.

CDER: exposure bridging, not immune-response endpoints

CDER's closest analogues are the anthrax antitoxin monoclonal antibodies approved under the Animal Rule. Their efficacy was shown on survival in rabbits and cynomolgus macaques. The human dose was justified by cross-species pharmacokinetic and exposure–response comparisons:

• Raxibacumab: At 40 mg/kg, human serum levels were comparable to concentrations associated with survival in animals ^179180181.
• ANTHIM (obiltoxaximab): At 16 mg/kg, humans were predicted to have similar peak concentrations and twice the overall exposure seen in infected animals at a fully effective dose ¹⁸⁴¹⁸⁵.

These programs used the concentration of the administered antibody, not a host immune response ¹⁷⁸.

Other CDER-relevant findings point the same way:

• Beyfortus (nirsevimab): The review mentions serum neutralizing antibodies in a cotton-rat rechallenge study, but cautions that it is unclear whether the finding can be extrapolated to humans. It does not present titers as an approval endpoint ³⁰⁶.
• Evusheld (COVID-19 pre-exposure prophylaxis, EUA): FDA's December 2021 announcement based the authorization on the placebo-controlled PROVENT trial, which showed a 77% lower risk of COVID-19. The announcement does not cite neutralizing titers ⁴⁸⁷⁴⁸⁸.
• Rabies mAb cocktails (draft guidance): FDA warns that rabies neutralizing antibody levels do not measure breadth of activity. It says the efficacy assumption applied to standardized human rabies immune globulin cannot be extrapolated to monoclonal antibody cocktails ²⁰⁸.

Practical takeaways for sponsors

• Choose the category up front. An established correlate (anti-HBs, hSBA, OPA for IPD) supports traditional approval. Immunobridging needs a reference population or product with demonstrated efficacy. An unvalidated marker leads to accelerated approval with a confirmatory study.
• Expect standard margins. The recurring immunobridging criteria are a GMT ratio lower bound of >0.67 (or an upper bound ≤1.5 when the ratio is inverted) and an SRR or seroconversion difference lower bound >−10 percentage points ^141339668674. Pneumococcal OPA GMT bridges have used a lower bound of >0.5 ¹⁵¹.
• Missing one co-primary endpoint matters. FDA reported the Moderna booster SRR miss explicitly ³⁴⁴. For PCVs, FDA looked to supportive immune data before accepting effectiveness for serotypes that missed ⁵⁰³⁵⁰⁷.
• Thresholds derived from animals carry visible uncertainty. FDA documents the gap between the animal endpoint (viremia, survival) and human disease, and it ties approval to confirmatory studies or Animal Rule conditions ^75658313.
• Accelerated approval on an immune marker needs a confirmatory study that can actually be completed. The IXCHIQ suspension shows the regulatory exposure when confirmatory trials are still incomplete and safety signals emerge ⁵⁸³⁵⁹¹.