The problem: inherit a route, then make it manufacturable
When Indivior in-licensed the selective orexin-1 receptor (OX₁R) antagonist program C4X3256 (INDV2000, rocavorexant) from C4X Discovery, it inherited a route that had delivered two early batches (480 g, then a 4.2 kg cGMP lot) at excellent purity — but with a list of scale-up liabilities: a telescoped Stage 1/2 of “complex equilibria,” a LiAlH₄ amide reduction, an mCPBA oxidation, a highly toxic TMSCN step, a structural-alert pyridine N-oxide (ICH M7), a corrosive SOCl₂ acid-chloride endgame, and high-molecular-weight salt intermediates that forced salt-break operations. This is the archetypal post-in-licensing problem: a chemistry that works is not yet a chemistry you can run.
The strategy: a convergent 10-stage route, two halves joined by T3P

The campaign kept the molecule’s convergent logic and rebuilt the operations around it. Two advanced intermediates are made in parallel and united in the final stage:
- Amine 6 — five stages from L-alaninamide·HCl: reductive amination (PhCHO), N-methylation, amide→amine reduction, S_NAr onto 2-chloro-5-(trifluoromethyl)pyrazine, then debenzylation.
- Acid 10 — four stages from 5-bromo-2-methylpyridine: Cu-catalyzed Ullmann N-arylation with 1,2,3-triazole (N2-regioselective), N-oxidation, a Reissert-type cyanation (TMSCN/BzCl) installing the C-2 nitrile, and hydrolysis.
- Final coupling — the two fragments joined with T3P / iPr₂NEt in EtOAc to give C4X3256 monohydrate in 80%.
The process intelligence: what actually changed
| Liability in the inherited route | Scale-up fix |
|---|---|
| Stage 1/2 telescope = “complex equilibria,” low/variable yield | Isolate crystalline amine 2 as a control point; revised Stage 1 (PhCHO/NaOH/Pd-C, MeOH) completes in 1–2 h |
| Poorly crystallizing amide intermediate 3 | Telescope Stages 2→3 in THF, avoiding its isolation |
| High-MW trimesic-acid salt of 6 forcing a salt break | Improved 6 and 10 so the final amide forms without a preliminary salt break |
| mCPBA N-oxidation | Peracetic acid |
| SOCl₂ acid-chloride endgame (corrosive, water-sensitive) | T3P coupling — equimolar acid 10 / amine 6, EtOAc, 35–45 °C |
The toxic steps that remain are handled, not hidden: the TMSCN/BzCl cyanation (HCN risk) runs cold with controlled addition, and all cyanide-containing waste is quenched, basified, and consigned for licensed disposal. The resulting N-oxide (ICH M7 alert) is carried as a transient, purged downstream.
The outcome
Two GMP batches delivered 14.4 kg (77%) and 14.8 kg (80%) — 29.2 kg total, ~78% average for the final stage — of C4X3256 monohydrate (Form 1) as white acicular needles, 99.7–99.8% HPLC and 100% ee. KF (3.8–4.0% w/w H₂O) confirmed the monohydrate (theory 4.1%); drying was capped at 50 °C to protect the hydrate. Tightening the crystallization concentration pulled D90 from 311 µm (prior lot) to 52 µm — a formulation-ready particle size out of the final step.
The drug substance itself: N,6-dimethyl-3-(2H-1,2,3-triazol-2-yl)-N-[(2S)-1-{[5-(trifluoromethyl)pyrazin-2-yl]amino}propan-2-yl]pyridine-2-carboxamide (anhydrous C₁₈H₁₉F₃N₈O, MW 420.4), isolated as the 1:1 hydrate.
Why it matters
- A textbook de-risking. New crystalline control points, two telescopes, and reagent swaps (peracetic acid for mCPBA, T3P for SOCl₂) turned an inherited medicinal-chemistry route into a kilogram-scale GMP process.
- Convergence pays. Building amine 6 and acid 10 separately and coupling last keeps the longest linear sequence short and the high-value fragments protected.
- Particle engineering at the API step. Hitting Form 1 monohydrate with a 52 µm D90 straight from crystallization is the kind of endgame control that formulation teams actually ask for.
Reference: Breen, G. F.; Carey, J. S.; Claveau, R.; Maczka, M. “Synthesis of a Selective Orexin 1 Receptor Antagonist: First Scale-Up Campaign Following In-Licensing.” Org. Process Res. Dev. 2026, DOI: 10.1021/acs.oprd.6c00444. (Work by Indivior and Almac Sciences; NIDA award UG3DA050308.)
Source article is CC-BY-NC-ND — no paper figures are reused here. Both visuals are original to Astinova Bio Labs: the convergent route and all structures redrawn with RDKit/Matplotlib.
Astinova Bio Labs develops and de-risks synthetic routes like this one — convergent design, reagent and safety swaps, telescoping, and crystallization/particle control. Talk to us about your process.