New Approach Methodologies in FDA Novel Drug and Biologic Approvals, 2022–2026

NAM use in FDA novel drug and biologic approvals, Nov 2022 to Aug 2026

Methodology

This survey covers 175 of the 196 original applications in Drugs@FDA coded as a new molecular entity (chemistry type 1 or 1/4) with action dates from November 17, 2022 to August 19, 2026 (115 NDAs, 60 BLAs). Each application's review package (pharmacology/toxicology, clinical pharmacology and multidisciplinary reviews) was searched for non-animal methods (in vitro phototoxicity, reconstructed-tissue irritation/sensitization, hiPSC and microphysiological systems, human-cell cytokine release and other human-cell assays, in silico/computational toxicology, weight-of-evidence (WoE) replacement of animal studies, and cases where no relevant animal species existed). Per-application findings were extracted with source citations. I then applied exclusion rules. Excluded were the routine ICH battery (Ames, in vitro clastogenicity, hERG, receptor panels), CYP/transporter in vitro DDI studies, PBPK, population PK and exposure-response modeling, QSAR impurity assessments, routine tissue cross-reactivity, and animal-study waivers based only on ICH S9 (advanced cancer), ICH S6 or single-dose use with no WoE analysis. FDA's March 2026 draft NAM guidance and related guidances were pulled for context. CBER-regulated products (cell and gene therapies, vaccines) are generally not in Drugs@FDA and are outside this survey.

Bottom line

• 126 of 175 novel approvals (72%) document at least one qualifying non-animal method. 109 have an actual non-animal assay or model (not counting WoE-only cases).
• The use is broad but shallow. Most of it is established, OECD/ICH-accepted in vitro testing: 3T3 NRU phototoxicity under ICH S10, BCOP and reconstructed human epidermis. That tier alone covers 63 approvals. FDA's draft NAM guidance names eye irritation, skin irritation and skin sensitization NAMs as already accepted as fit-for-purpose ⁷¹⁹.
• Advanced NAMs are rare. Complex in vitro systems (MPS, 3D liver spheroids, hiPSC-derived tissues) appear in only 5 approvals. Computational toxicology models used for safety decisions appear in 13.
• The only organ-on-chip/MPS case is NEREUS (tradipitant). The applicant offered liver-chip, 3D hepatocyte spheroid and four-organ MPS data as alternative evidence against a chronic nonrodent study. FDA found that these data could not support regulatory decision-making ^145214511458.
• Non-animal data replaced an animal study mostly through WoE assessments, not through a new assay. The pattern is literature- and mechanism-based carcinogenicity or DART waivers for monoclonal antibodies and other biologics, often with Executive Carcinogenicity Assessment Committee (ECAC) concurrence. In most other cases, the reviews describe in vitro human data as supplementing animal studies rather than replacing them.

Regulatory context

• FDA's draft guidance General Considerations for the Use of New Approach Methodologies in Drug Development (March 18, 2026) defines NAMs as complex in vitro, 2D in vitro, in chemico and in silico methods. It sets out a validation framework built on context of use, human biological relevance, technical characterization and fit-for-purpose ^717718722723.
• Section 3209 of FDORA (Consolidated Appropriations Act, 2023) amended FD&C Act 505 and PHS Act 351 to clarify that nonanimal testing data may support an IND or biosimilar BLA ⁷¹⁷.
• The draft guidance Monoclonal Antibodies: Streamlined Nonclinical Safety Studies (December 2, 2025) invites sponsors to propose NAMs and reduced-animal designs ⁷³⁰. The CGT FAQ guidance (August 19, 2026) states FDA support for NAMs and the 3Rs ⁷²⁵.
• Almost all approvals in this window had nonclinical programs designed before these documents were issued. The low uptake of advanced NAMs is therefore expected and should be read as a baseline.

Method tiers used

Clustering by therapeutic area

Therapeutic areas were assigned from each product's labeled indication. Where the review text did not state it, I used the indication pulled from the label.

Observed patterns

• Oncology has the most NAM use by volume. Bispecific T-cell engagers, PD-1/PD-L1 antibodies and ADCs account for most human cytokine-release assays. Oncology biologics also account for most WoE-based DART waivers. For small-molecule kinase inhibitors, NAM use is mostly 3T3 NRU phototoxicity testing. Two of the five complex in vitro cases are oncology products: BIZENGRI (hiPSC cardiomyocyte viability) ⁹³⁰ and HYRNUO (ReproTracker hiPSC teratogenicity assay) ¹²⁷³.
• Immunology, dermatology, non-malignant hematology and ophthalmology monoclonal antibodies cluster in WoE carcinogenicity waivers. These were typically needed because the antibody was not pharmacologically active in rodents: NEMLUVIO ⁷⁷²⁹⁰⁸, OMVOH ¹²⁴⁷, EBGLYSS ¹²⁵⁵, BIMZELX ¹¹⁰⁰, PIASKY ⁹¹⁸, ALHEMO ¹⁰⁰⁶, LUMVOA ⁷⁷⁵.
• Oligonucleotides (5 of 6 had a qualifying method) cluster in human-hepatocyte target-engagement assays and in silico off-target transcriptome analyses. Examples: TRYNGOLZA ^17061690, DAWNZERA ^12961299, WAINUA ¹⁷⁷², QFITLIA ^17101717, REDEMPLO ¹⁴²¹.
• Cardiac and mitochondrial disease accounts for the hiPSC-cardiomyocyte cases with a mechanistic or efficacy rationale: MYQORZO engineered heart tissues from MYH7 R403Q donors ¹⁷³⁹, and FORZINITY Barth syndrome iPSC-cardiomyocytes ¹⁹⁸⁵. The FORZINITY reviewer considered the evidence limited.
• Topical dermatology and ophthalmic small molecules cluster in local-tolerance NAMs (BCOP, RhE, UV absorbance waivers).
• Psychiatry, pain and diagnostic imaging show the least use beyond routine phototoxicity or single-dose waivers.

Clustering by modality

Modality comes from the review text. "Other/NS" means the retrieved text did not state the modality.

Tier 1: complex in vitro systems (MPS, 3D, hiPSC)

• MYQORZO (Cardiovascular/lipid/obesity, approved 2025-12-19): Engineered heart tissues composed of induced human pluripotent stem cell-derived cardiomyocytes from wild-type and MYH7 R403Q-heterozygous donors were used; aficamten reduced contractility in both tissues. ¹⁷³⁹ Role: Supplemented animal and other nonclinical pharmacology studies and supported the proposed mechanism of action and efficacy rationale in HCM-relevant human cellular tissue; the text does not state that it replaced an animal study. ^{1739 1739}
• NEREUS (Hepatology/GI, approved 2025-12-30): Three-dimensional primary human-hepatocyte spheroid studies (including studies with tradipitant and metabolites, and co-exposure to cytokines or bile acids) were conducted to quantify in-vitro hepatotoxic potential and explore mechanisms. ^{1444 1447} Role: Supplemented animal toxicology by assessing whether hepatotoxicity could occur in human donor-derived liver cells. The Applicant included these assays in a totality-of-evidence argument that a chronic nonrodent study was not justified; FDA concluded that the additional in-vitro assays did not adequately support that conclusion. ^{1451 1444}
• NEREUS (Hepatology/GI, approved 2025-12-30): A human co-culture Liver-Chip (liver MPS) comprising primary human hepatocytes and liver sinusoidal endothelial cells under flow was used to evaluate tradipitant and metabolite toxicity for up to 7 days. ¹⁴⁵⁸ Role: Supplemented animal studies with a human liver-relevant toxicity model and supported evaluation of whether liver toxicity was animal-specific. It did not replace animal testing; FDA considered the studies of limited utility for clinical-risk assessment. ^{1461 1458}
• NEREUS (Hepatology/GI, approved 2025-12-30): Two four-organ human microphysiological-system studies used primary human hepatocytes, iPSC-derived cardiomyocytes, primary human renal proximal-tubular epithelial cells, and iPSC-derived cortical neurons to prospectively evaluate tradipitant safety. ¹⁴⁵² Role: Intended by the Applicant as pivotal alternative evidence to support safety and, together with other in-vitro studies, its position that a chronic nonrodent toxicity study was not justified. FDA found the studies did not substantively contribute to the safety profile and could not support regulatory decision-making. ^{1452 1452}
• BIZENGRI (Oncology/heme-onc, approved 2024-12-04): In vitro cardiomyocyte-viability assay using adult human stem-cell-derived cardiomyocytes in the presence of doxorubicin; zenocutuzumab did not affect viability. Role: Supplemented the cardiac safety assessment; supported the conclusion of no cardiomyocyte-viability effect and did not replace an animal study.^{930 930}
• HYRNUO (Oncology/heme-onc, approved 2025-11-19): ReproTracker human-induced pluripotent stem-cell (hiPSC) assay assessed differentiation into cardiomyocytes, hepatocytes, and neural rosettes. Sevabertinib disrupted cardiomyocyte and hepatocyte differentiation and was identified as a teratogen by the assay criteria. Role: Supplemented the rat embryo-fetal-development study and the reproductive-risk weight-of-evidence assessment; supported reproductive-risk characterization and embryo-fetal toxicity/contraception recommendations. It was not described as replacing the rat study. ¹²⁷³
• FORZINITY (Rare genetic/metabolic, approved 2025-09-19): Elamipretide (1 µM) was evaluated for effects on oxygen-consumption rate in induced pluripotent stem cell-derived cardiomyocytes from subjects with BTHS. The applicant reported a qualitative improvement under maximal-respiration conditions, although the reviewer noted that primary data and a peer-reviewed publication were unavailable. Role: Supplemented animal BTHS-model data as a human-relevant mechanistic/proof-of-concept assessment; it did not replace or reduce an animal study, and the reviewer considered the evidentiary support limited. ¹⁹⁸⁵

Tier 2: computational and in silico safety models

• FOUNDAYO (Cardiovascular/lipid/obesity, approved 2026-04-01): FDA NCTR applied the rule-of-two (RO2) and DILIscore computational models to evaluate potential human drug-induced liver injury risk. Role: Supplemented animal and other nonclinical findings with a human DILI-risk assessment; supported evaluation of hepatotoxicity risk, not replacement of an animal study. ¹¹³⁴
• VEOZAH (Endocrine/bone, approved 2023-05-12): FDA’s nonclinical team consulted the Division of Applied Regulatory Science Computational Toxicology Consultation Service and performed an evaluation of structural alerts for DNA reactivity, carcinogenicity signals in nonclinical bioassays, and potential mechanisms of tumorigenesis/tumor-promotion activity for fezolinetant and its metabolites. ¹⁷⁸⁸ Role: Supplemented—not replaced—the rat and transgenic-mouse carcinogenicity studies; supported the conclusion that fezolinetant was not associated with increased tumor-promotor activity after a clinical-trial malignancy imbalance was identified. ^{1788 1788}
• YORVIPATH (Endocrine/bone, approved 2024-08-09): Quantitative structure-activity relationship (QSAR) analysis of TransCon Linker structures, including structures with and without the mPEG branching point, found no genotoxic structural alerts or alerts for hepato-, renal-, cardio-, developmental/reproductive, skin-irritation, systemic, or phototoxicity hazards. Role: Supplemented the nonclinical hazard assessment for the linker; it supported a negative structural-alert assessment and was not described as replacing an animal study. ²⁰⁴⁹
• YORVIPATH (Endocrine/bone, approved 2024-08-09): In silico analysis of the palopegteriparatide linker found no structural alerts for cardiotoxicity, including QT-interval prolongation. Role: Supplemented cardiovascular safety assessment supporting the conclusion that hERG inhibition was unlikely; no animal-study replacement was stated. ²⁰⁵⁰
• YUVIWEL (Endocrine/bone, approved 2026-02-27): An in silico evaluation identified no genotoxicity risk. This evaluation was also included among the evidence considered in the carcinogenicity weight-of-evidence assessment. Role: Supplemented the nonclinical evidence base supporting the conclusion of no genotoxicity risk and the carcinogenicity risk assessment; it did not itself replace animal testing. ¹⁴¹⁵
• QFITLIA (Hematology (non-malignant), approved 2025-03-28): Cheminformatics assessment and in silico modeling by FDA's Division of Applied Regulatory Science of fitusiran hepatotoxicity potential relative to marketed siRNA products, based on chemical structure. Role: Supplemented the clinical pharmacology/DILI assessment of hepatotoxicity potential; no animal study replacement is stated. ¹⁷¹⁷
• VOYDEYA (Hematology (non-malignant), approved 2024-03-29): Computational amino-acid sequence homology, phylogenetic-tree, and sequence-alignment analyses of MT1 receptors from humans, dogs, rabbits, and rats. The analyses supported dog and rabbit as pharmacologically relevant species for assessing potential off-target MT1 effects. Role: Supplemented the animal-species selection rationale for toxicology studies; no animal-study replacement or reduction was stated. ¹⁶³¹
• NEREUS (Hepatology/GI, approved 2025-12-30): In-silico toxicity analyses of tradipitant and metabolites were conducted, including updated machine-learning-assisted structural/toxicity predictions using a Grok-based workflow, RDKit descriptors, qualitative consensus-classifier thresholds, and in-house Python scripts. ¹⁴⁶⁰ Role: Supplemented the assessment of metabolite safety and structural toxicity risks; it did not replace an animal study. FDA considered the broader in-silico findings not toxicologically relevant in light of completed in-vitro and in-vivo toxicity studies. ^{1452 1460}
• DAWNZERA (AUTOINJECTOR) (Immunology/allergy, approved 2025-08-21): In silico and in vitro characterization of PKK-ASO specificity was conducted. The review states that the unconjugated ASO with the same sequence was perfectly matched only to PKK RNA in the human transcriptome, supporting sequence/target specificity. Role: Supplemented nonclinical pharmacology evidence; supported the assessment of molecular target specificity rather than replacing an animal study. ¹²⁹⁶
• IMAAVY (Neurology, approved 2025-04-29): Computational protein-sequence comparisons assessed potential nipocalimab cross-reactivity with human proteins and FcRn from selected nonhuman mammalian species. A genome aggregation database analysis assessed FcRn coding variants at the nipocalimab-FcRn binding interface. Role: Supplemented specificity, species-relevance, and genetic-variant assessments. These analyses supported interpretation of off-target potential and FcRn binding-interface variability; the text does not state that they replaced or reduced animal studies. ⁸⁵²
• BIZENGRI (Oncology/heme-onc, approved 2024-12-04): In silico immunogenicity screening assessed binding of zenocutuzumab-derived 10-mer peptides to human HLA class II molecules; results were comparable to marketed humanized benchmark antibodies. Role: Supplemented immunogenicity risk assessment, supporting the conclusion that zenocutuzumab had no increased immunogenicity risk versus benchmark humanized antibodies; it did not replace an animal study.^{930 930}
• XDEMVY (Ophthalmology, approved 2023-07-24): In silico Derek Nexus evaluation for respiratory or skin sensitization; no alert was triggered for respiratory or skin sensitization. Role: Supplemented sensitization hazard assessment; absence of skin sensitization was subsequently confirmed in a mouse local lymph node assay, so it did not replace the animal study. ¹⁹²⁹
• TRYNGOLZA (AUTOINJECTOR) (Rare genetic/metabolic, approved 2024-12-19): Sequence-dependent off-target assessment used in silico predictions to identify potential off-target transcripts, expression-data resources (GTEx and Protein Atlas) to assess liver expression, Ingenuity Pathway Analysis to assess associations with liver-toxicity endpoints, and an additional BLASTn/UCSC-browser sequence search. Predicted RAC1 and STIM2 off-target effects were also evaluated in cultured primary human hepatocytes by RT-qPCR. Role: Supplemented the hepatic safety assessment by evaluating the potential for sequence-dependent off-target effects relevant to liver toxicity; it supported the conclusion that there was no suggestion of relevant on-target or nonspecific off-target mRNA effects. It was not described as replacing an animal study. ¹⁷⁰⁶
• EXDENSUR (Respiratory, approved 2025-12-16): An in silico placental-transfer model was developed to compare monkey and human distribution/fetal exposure profiles for depemokimab versus mepolizumab. It predicted fetal PK differences associated with depemokimab's YTE modification; the model was verified against monkey toxicology PK data and mepolizumab ePPND PK data. Role: Supplemented—not replaced—animal evidence. It supported assessment of YTE-modification effects on placental transfer/fetal exposure and informed pregnancy labeling; residual uncertainty led to a postmarketing pregnancy safety study requirement. ⁸³⁰

Three of these were run by FDA itself, not the applicant: VEOZAH (Division of Applied Regulatory Science computational toxicology consult on tumor-promotion signals) ¹⁷⁸⁸, FOUNDAYO (NCTR rule-of-two and DILIscore liver-injury models) ¹¹³⁴, and QFITLIA (DARS cheminformatics hepatotoxicity comparison against marketed siRNAs) ¹⁷¹⁷.

Cases where a non-animal approach replaced, reduced or avoided animal work

• UNLOXCYT: Cosibelimab did not bind mouse PD-L1, so human in vitro assays (reporter, MLR, ADCC) replaced in vivo mouse efficacy studies ¹²³².
• PIASKY: Drug-target-drug complex formation was studied in vitro with human C5 and plasma instead of in animals ⁹¹⁸.
• APHEXDA: In vitro human decidual, trophoblast and neural-precursor data plus mechanism supported the conclusion that in vivo reproductive toxicity studies would add no useful information ¹⁹¹⁴.
• ZELSUVMI / IDVYNSO / SOFDRA: UV-visible absorbance or molar extinction below the ICH S10 threshold made further phototoxicity testing unnecessary ^196219681769.
• TALVEY: No pharmacologically relevant species existed. FDA agreed that more monkey and rodent studies were not warranted, and a surrogate tool molecule was used for limited hazard identification ^978979977.
• LUMVOA / LEROCHOL: In vitro binding showed no rodent activity, so a single species (cynomolgus) was used and rodent carcinogenicity was replaced by WoE ^775766767.
• ZILBRYSQ: Comparative human and monkey hepatocyte metabolism supported a one-species toxicology program ¹⁹⁰¹.
• LIFYORLI: Human versus dog primary-hepatocyte GR assays showed the dog was an unsuitable toxicology species ¹⁰⁷⁴.
• LOARGYS: In vitro proliferation data fed a WoE carcinogenicity assessment that ECAC accepted ¹¹⁴⁹.

Tier 3: validated in vitro photo and local toxicity

In most 3T3 NRU cases, a negative in vitro result was used under ICH S10 to conclude that no in vivo phototoxicity study was needed. Where 3T3 NRU was positive (for example, the DXd payload of DATROWAY), a pigmented-rat in vivo study followed ⁸³⁵⁷⁸⁶.

Tier 4: human blood and cell safety assays

The reviews generally describe these assays as supplementing animal toxicology. There were two notable decision uses. For ENFLONSIA, human whole-blood cytokine data supported lifting a partial clinical hold ⁷⁶³. For LUNSUMIO VELO, PBMC EC20 values informed the first-in-human starting dose ¹¹¹⁰.

Tier 5: human-cell pharmacology and species-relevance work

Tier 6: weight-of-evidence replacement of animal studies

WoE-based waivers are a 3Rs mechanism, not a NAM in the narrow assay sense. They are kept as a separate tier because FDA's draft NAM guidance lists support for WoE approaches as a context of use for NAMs ⁷¹⁸.

Limitations

• The survey covers November 2022 to August 2026: 175 of the 196 novel approvals with action dates in that window (about 89%). The other 21 were not reviewed, and approvals before November 17, 2022 are outside its scope.
• CBER products (cell and gene therapies, vaccines, blood products) are not covered.
• Findings depend on what reviewers wrote. A method the applicant ran but the review did not describe would be missed. Extraction was automated per application and then filtered by rule, so borderline items (for example, human-cell pharmacology) may be over- or under-counted.
• PBPK (terms found in reviews for 84 of 175 applications) and QSAR impurity assessments (23 of 175) are in silico NAMs under a broad definition. They were excluded because they are routine and outside the toxicology-replacement question.
• Therapeutic area and modality assignments are based on keywords and are approximate.