Integrated vs Standalone ICH S7A Safety Pharmacology in FDA-Approved Small-Molecule NMEs: Oncology vs Non-Oncology Precedents

Safety pharmacology integrated into toxicology vs standalone: FDA NME small-molecule NDAs, Sep 2021 to Sep 2026

Scope and method

• Universe: Original NDAs with Drugs@FDA chemistry type 1 or 1/4 (NME) and action dates from 2021-09-28 to 2026-09-28. That gave 168 applications. After removing peptides, oligonucleotides, aptamers, imaging and contrast agents, radiopharmaceuticals, gases, and botanical or polymer products, 124 small-molecule NME products were left.
• Source: For each product, the FDA pharmacology/toxicology review in Drugs@FDA was queried for how the ICH S7A core battery (CNS, in vivo cardiovascular, respiratory) was assessed. Each system was classified as a dedicated standalone study (S), endpoints built into general or repeat-dose toxicology (I), both (B), or not conducted/waived (W). Products with gaps got a second pass using different search terms.
• Classes:
  • Fully integrated: no dedicated in vivo core-battery study was described, and every system with data was covered in toxicology studies.
  • Hybrid: at least one system was covered only in toxicology studies, and at least one had a dedicated study.
  • Standalone battery: every system with data had a dedicated study. ECG monitoring in nonrodent toxicology studies may also have been done (coded B).
  • Insufficient: fewer than two systems could be characterized from the posted documents.
• Caveat: The classifications come from machine-assisted reading of review excerpts, not a full manual read of each review. "Standalone" means the pages retrieved described dedicated studies. It does not rule out extra endpoints in toxicology studies that the excerpts did not mention.

Headline numbers

• Of the 94 products that could be characterized, 21 (22%) replaced at least one dedicated core-battery study with endpoints in toxicology studies.
• The split by indication:
  • Oncology: 11 of 28 characterized products (39%).
  • Non-oncology: 10 of 66 (15%).
• Cardiovascular is almost never merged. 85 of the 94 characterized products had a dedicated in vivo cardiovascular study, usually telemetry in dog or monkey. In 28 of those, ECG was also collected in repeat-dose toxicology studies. CNS was integrated in 19 products and respiratory in 14.
• The typical hybrid design pairs a standalone nonrodent telemetry study with CNS observations (Irwin/FOB) and respiratory measures built into the 28-day or 13-week rat toxicology study.

Regulatory framework (why both designs are acceptable)

• ICH S7A: Safety pharmacology endpoints may be built into toxicology studies. Adequately designed toxicology studies can reduce or eliminate separate studies, and incorporated studies would be GLP ¹⁷⁴. S7A also cautions that routine toxicology observations may miss effects that a dedicated study would detect ¹³⁵⁹. Separate studies may not be needed for locally applied agents with low systemic exposure, or for cytotoxics in end-stage cancer ¹³⁵⁸.
• ICH M3(R2): Recommends adding in vivo safety pharmacology evaluations to general toxicity studies "to the extent feasible" to reduce animal use ¹⁷².
• ICH S9: The core battery may be built into general toxicology studies. Detailed clinical observations plus appropriate nonrodent ECG are generally enough. Standalone studies are not routinely called for in advanced cancer unless a specific concern exists ^14321436.
• ICH S7B and E14/S7B Q&As: Toxicology-study ECGs are treated as supporting information. The in vivo QT assay (customarily conscious telemetered animals) carries specific sensitivity and exposure expectations ^137914031404. This explains why cardiovascular stays standalone even when CNS and respiratory are merged.
• Oncology radiopharmaceuticals (FDA guidance): Standalone cardiovascular, respiratory and CNS studies are generally not warranted ¹⁸¹. Rare Diseases (FDA guidance): In vivo safety pharmacology can generally be integrated into general toxicity studies ¹⁷⁶.

When merging happened: drivers seen in the reviews

1. Oncology indication (ICH S9). This covers most integrated and hybrid products. Asciminib, pacritinib, adagrasib, quizartinib, inavolisib, taletrectinib, vepdegestrant, sonrotoclax, vorasidenib, dordaviprone and imlunestrant all merged some or all endpoints. For capivasertib, FDA explicitly tied the incorporated endpoints (clinical observations, ECG) to ICH S9 ⁶⁵⁶. For sevabertinib, the review likewise notes that CNS observations in the 4-week rat study were "concordant with ICH S9" ³⁷⁵. However, 17 of the 28 characterized oncology NMEs still ran a full standalone battery. Examples: elacestrant ¹²⁴⁶, momelotinib ¹⁰⁸⁰, fruquintinib ⁹⁸⁶, nirogacestat ⁸¹⁷, lazertinib ⁶⁵⁴, ensartinib ⁸⁰³, vimseltinib ²⁷², revumenib ⁶⁸⁸. S9 allows integration but does not require it.
2. Species relevance. For avacopan, the rat CNS and respiratory studies were in a pharmacologically non-relevant species. FDA recommended adding those evaluations to the chronic monkey toxicology study ¹¹⁸⁴. Suzetrigine placed FOB in the 28-day monkey toxicity study alongside a dedicated monkey safety pharmacology study ²¹⁹. Sepiapterin used marmoset repeat-dose studies for ECG ³⁷⁹.
3. Local or nonsystemic routes. For lotilaner, FDA told the sponsor safety pharmacology could be submitted as standalone studies or as endpoints in GLP ocular toxicity studies ⁸²². Ensifentrine assessed CNS with a modified Irwin test inside a 28-day inhalation toxicity study ⁹⁸⁸. For aceclidine, FDA's advice depended on systemic exposure: safety pharmacology endpoints could be built into the chronic ocular study under those conditions ⁶¹⁴⁶¹⁶.
4. Program-level choices in non-oncology. Deucravacitinib is the clearest non-oncology example of a fully merged battery. Neurobehavioral and respiratory assessments sat in the 1-month rat and monkey studies, and cardiovascular ECG sat in the 9-month monkey study ¹¹¹⁸. Zavegepant ⁹⁸¹, danicopan ⁷⁹⁵ and gepirone ¹¹²⁶ merged CNS and respiratory but kept or added dedicated cardiovascular work.
5. Omission justified by toxicology data. For palovarotene, no respiratory safety pharmacology study was run, and FDA relied on the absence of respiratory findings in repeat-dose toxicity studies ⁹⁷⁸. For deuruxolitinib, FDA agreed additional CNS, cardiovascular and pulmonary studies were not needed ⁵¹⁸.

Fully integrated (no dedicated in vivo core-battery study described)

Hybrid (at least one system covered only in toxicology studies)

Standalone battery (dedicated studies for all systems with data)

In a subset, ECG was also collected in repeat-dose toxicology studies (CV coded B). In capivasertib, sevabertinib and gedatolisib, endpoints were duplicated in both designs.

Insufficient review detail (30 products)

These products could not be characterized. Two things drove this: pharmacology/toxicology reviews not yet posted for many 2025 to 2026 approvals, where only labeling was indexed, and retrieved pages that covered only clinical QT or general toxicology.

Vtama ¹¹⁸³, Relyvrio ⁹⁸⁹, Lytgobi ¹⁵²⁴, Rezlidhia (monkey ECG in 28-day study only described) ⁹⁷², Sunlenca ¹⁰⁴³, Jaypirca ⁸⁶⁹, Jesduvroq ⁹³⁷, Joenja ⁵¹⁷, Paxlovid (nirmatrelvir telemetry only) ¹³⁵⁵, Inpefa ¹⁰⁵⁵, Augtyro (91-day monkey ECG only) ⁷⁷⁷, Vafseo ¹¹¹¹, Ojemda ⁷⁸⁰, Sofdra (minipig tox ECG only) ¹²⁵⁵, Cobenfy ¹⁴⁹⁸, Alyftrek ⁸⁰², Tryptyr ¹⁰⁰⁹, Ekterly ²¹⁸, Vizz ⁶¹⁴, Lynkuet ⁵⁰³, Komzifti ⁴⁶¹, Nereus ³¹⁰, Adquey ⁴⁶⁰, Bysanti (relied on iloperidone data via PK bridge) ⁴⁰⁷, Foundayo ¹⁹⁴, Orzeyful ⁴¹⁷, Zenbexus ²⁰⁹, Rasonque ³³⁹, Lisraya ⁴²⁹, Etcamah ²⁷⁰.

Scope exclusions to be aware of

Peptides, oligonucleotides and diagnostic agents were excluded by design. Trofinetide, levacetylleucine and taurolidine (Defencath) were also excluded from this pass and were not assessed. The oncology flag follows indication: mirdametinib (NF1 plexiform neurofibromas) was counted as non-oncology.