Human vs Cynomolgus Antigen Selection

Every BsAb program inherits a species question: which ortholog of each target do we express, screen against, and carry into tox? Cynomolgus macaque (cyno) is the default non-human primate for biologics, but it is not free — cyno reagents cost time and money, and cross-reactivity is never guaranteed by sequence homology. These decisions are per-target, per-arm, and should be made deliberately, not by default.

Decision 1 — Do you need cyno at all?

Choose cyno as mandatory when the molecule needs an NHP tox study and no rodent surrogate can carry it: T-cell engagers with a CD3 arm, checkpoint/co-stimulatory immunomodulators (e.g., PD-1), and Fc-engineered effectors whose ADCC/cytokine biology must be read in a relevant species. The FDA review for pembrolizumab is the canonical rationale — cyno was selected as the tox species on the basis of PD-1 sequence homology plus binding affinity ([FDA CrossR](https://www.accessdata.fda.gov/drugsatfda_docs/nda/2014/125514Orig1s000CrossR.pdf)). Choose cyno when the CD3 arm is the design driver: anti-CD3 discovery campaigns routinely clone and screen both human and cyno CD3ε, because reported ~96% identity does not guarantee binding — cyno CD3ε carries reported polymorphic epitopes (the FN18 polymorphism) that change antibody recognition ([Liu & Wang](https://onlinelibrary.wiley.com/doi/abs/10.1038/sj.icb.7100042), [US20230357398](https://patents.justia.com/patent/20230357398)).

Choose human-only when you are still in human-first discovery, when the target is conserved and cross-reactive in rodents and a rodent tox strategy is regulator-acceptable, or when the current milestone is binding/developability on the human antigen. Cyno is optional, never automatic: defer it until a tox-species decision exists.

Decision 2 — Cross-reactivity screening strategy

Choose to screen both directions once a lead format is fixed: human binder against cyno ortholog, and cyno binder against human antigen (relevant for cyno-derived or cyno-selected binders). Run recombinant ECD binding (SPR/BLI KD on human vs cyno ECD) *plus* cell-based binding (cyno PBMC or ortholog-overexpressing cells) — ECD-only data miss epitope presentation and polymorphism effects.

Choose an acceptance window before you test: typical programs tolerate a human↔cyno KD shift of ≤2–5× for a single tox molecule; reported shifts >5–10× mean cyno PK/tox readouts lose quantitative value — decide then whether to engineer cross-reactivity deliberately or run a separate cyno-surrogate binder. Choose matched ortholog pairs once the tox species is confirmed; choose cyno-primary reagents only for arms where cyno biology drives the design (e.g., CD3-arm affinity attenuation).

Decision 3 — PK/PD and tox-species translation

Choose a cyno-cross-reactive lead when the cyno tox study must also generate PK/PD: the same molecule in tox and pharmacology removes surrogate-binder translation error. Choose cyno for PD readouts (T-cell activation, cytokine release/CRS) when the program is a TCE or checkpoint asset — rodent surrogate TCEs are reported to misestimate CRS magnitude, and cyno tissue-expression plus tox data underpinned tarlatamab's dossier ([EMA Imdelltra EPAR](https://www.ema.europa.eu/en/documents/assessment-report/imdylltra-epar-public-assessment-report_en.pdf)).

Do not over-promise: any human↔cyno affinity gap shifts exposure–response scaling and widens safety-margin uncertainty in human dose projection; a >5–10× KD gap makes cyno PK/PD modeling approximate, and cyno PK never guarantees human PK.

Decision 4 — Sequencing: human-first discovery, cyno later

Choose human-first discovery: primary screening against human antigens; cyno cross-reactivity is a secondary panel at lead optimization, not part of initial discovery — it protects reagent spend and keeps campaigns human-centric. Choose to run the human↔cyno ECD alignment *in silico* early to flag epitope residues that differ (CD3ε residues reported polymorphic in cyno are the classic case); if the epitope includes non-conserved residues, decide deliberately whether to engineer cross-reactivity. Choose cyno work only when the tox species is confirmed and the format is fixed (TCE geometry changes cyno PK demands) — i.e., IND-enabling stage, not before.

Tool set this implies (Human/Cyno matched pairs)

Reagent implication, per target per format: a matched pair of human ECD + cyno ECD reagents (Fc-tagged and biotinylated for SPR/BLI), ortholog-overexpressing cell lines with knockout controls, and cyno PBMC/whole blood for CD3-arm and cytokine-release assays. Golden Pool examples: PD-1 (reported ~87–90% human/cyno identity; cyno PD-1 is a catalog item ([AcroBiosystems](https://qr.acrobiosystems.cn/data/ds/PD1-C5253.pdf), [Creative Biolabs](https://www.creative-biolabs.com/nhp-biologicals/pdf/NHFF-0524-HX407.pdf))); CD3ε (reported ~96% identity but polymorphic — order the cyno CD3ε pair before committing the CD3 arm); EGFR/MET (reported >95% ECD identity; amivantamab ran cyno tox ([EMA Rybrevant EPAR](https://www.ema.europa.eu/en/documents/assessment-report/rybrevant-epar-public-assessment-report_en.pdf))); DLL3 (cyno tissue-expression was assessed in tarlatamab's dossier ([EMA Imdelltra EPAR](https://www.ema.europa.eu/en/documents/assessment-report/imdylltra-epar-public-assessment-report_en.pdf))).

LoFly D6 stance — cyno is P2, not P1

Cyno matched pairs are an IND-enabling add-on (P2), never the discovery backbone (human reagents = P1) — and we do not over-promise: cyno cross-reactivity does not guarantee tox predictivity, and every homology/KD statement above is reported/typical until your own SPR confirms binding on your molecule.

References

• FDA Pharmacology/Toxicology review, pembrolizumab NDA 125514 (cyno species justification): https://www.accessdata.fda.gov/drugsatfda_docs/nda/2014/125514Orig1s000CrossR.pdf
• Liu & Wang, "Polymorphisms of CD3ε in cynomolgus and rhesus monkeys and their relevance to anti-CD3 antibodies and immunotoxins," Immunology & Cell Biology: https://onlinelibrary.wiley.com/doi/abs/10.1038/sj.icb.7100042
• EMA EPAR, Imdelltra (tarlatamab, DLL3×CD3): https://www.ema.europa.eu/en/documents/assessment-report/imdylltra-epar-public-assessment-report_en.pdf
• EMA EPAR, Rybrevant (amivantamab, EGFR×MET): https://www.ema.europa.eu/en/documents/assessment-report/rybrevant-epar-public-assessment-report_en.pdf
• AcroBiosystems, cynomolgus PD-1/CD279 datasheet (vendor technical note): https://qr.acrobiosystems.cn/data/ds/PD1-C5253.pdf
• Creative Biolabs, cynomolgus PD-1 NHP biologicals datasheet (vendor technical note): https://www.creative-biolabs.com/nhp-biologicals/pdf/NHFF-0524-HX407.pdf
• US20230357398, "Novel human antibodies binding to human CD3 epsilon" (human + cyno CD3ε cloning in discovery): https://patents.justia.com/patent/20230357398