Patenting the Half-Built Molecule Chugai Claims Four Indole Fragments, Not the Drug

The title of this application promises a pyrazolopyridine derivative with GLP-1 receptor agonist effect. The claims deliver something quite different. All five claims are directed at four specific intermediate compounds, labelled 31f, 31i, 31j and 31k in the specification, none of which contains a pyrazolopyridine ring and none of which is reported to have any activity at the GLP-1 receptor at all. They are unfinished pieces: the indole half of a drug molecule, caught at four points along a nine step synthesis. AU 2026202206 is a divisional of AU 2024201884 filed by Chugai Pharmaceutical, and it was published on the same day as its sibling AU 2026202208, which claims the broad compound formula this one deliberately leaves alone.

The Problem

The specification body is shared across the whole family and frames the usual problem: every commercialised GLP-1 analogue at the 2016 priority date required an injection, so a non-peptide agonist that could be swallowed was worth chasing. The background runs through peptide workarounds such as pairing semaglutide with the absorption enhancer SNAC, and earlier small molecule attempts in WO 2009/111700 and WO 2010/114824 which it says still needed improvement in activity, metabolic stability and bioavailability.

But that is the problem the sibling application answers. The problem this divisional answers is narrower and more commercial. The lead molecules in this family are large, heavily substituted, and made by convergent routes in which two complicated halves are built separately and then joined by an amide coupling. If you own the formula covering the finished drug but not the pieces, a competitor can make and sell the pieces, and can do so in a country where the compound claim does not run. Claiming the intermediates closes that gap. It is a well worn move in pharmaceutical drafting and it is what the claims of AU 2026202206 are for.

There is a second, more mundane problem visible in the chemistry. The indole fragment carries three stereocentres and an acidic heterocycle, and the acid group that eventually forms the amide bond cannot simply be carried through the route unprotected. The specification’s answer is to run the whole sequence with the acid masked as an N-methyl-N-phenyl amide and only unmask it at the last moment. Three of the four claimed compounds carry that mask.

What This Invention Does

Each of claims 1 to 4 is a bare compound claim of the simplest possible kind: a drawn structure, then the words “or a salt thereof”. No activity, no use, no composition.

Claim 1 covers Compound 31f, 5-[(4S)-2,2-dimethyloxan-4-yl]-N-methyl-N-phenyl-1H-indole-2-carboxamide. Step 31-4 of the specification makes it by dissolving 805 mg (2.95 mmol) of the preceding indole-2-carboxylic acid in 8.0 mL of dimethylacetamide, adding thionyl chloride (0.256 mL, 3.53 mmol) dropwise at an internal temperature of 10 degrees C or lower, stirring for an hour, then adding N-methylaniline (0.384 mL, 3.53 mmol) and triethylamine (0.985 mL, 7.07 mmol) at the same temperature. Water is added, the solid is filtered off and dried, and the yield is 995 mg, or 93 per cent. Mass spectrometry gives m/z 363. The step before it, a sodium hydroxide hydrolysis at 65 degrees C, had run at 96 per cent.

Claim 2 covers Compound 31i, 1-[(1S,2S)-1-cyano-2-methylcyclopropyl]-5-[(4S)-2,2-dimethyloxan-4-yl]-N-methyl-N-phenylindole-2-carboxamide, at m/z 442. Getting there from 31f takes two steps. The indole nitrogen is first alkylated with 2-chloroacetonitrile (0.026 mL, 0.414 mmol) using 8 molar potassium hydroxide (0.103 mL, 0.828 mmol) in 1,3-dimethyl-2-imidazolidinone at an external temperature of 10 degrees C for 2.5 hours, on a 101 mg (0.276 mmol) scale. That gives the cyanomethyl compound 31h at m/z 402, which the specification says was taken straight into the next step as a light brown oil without purification. The cyanomethyl group is then closed into the cyclopropane ring to give 31i.

Claim 3 covers Compound 31j, at m/z 501, in which the nitrile of 31i has been converted into a 5-oxo-4H-1,2,4-oxadiazole ring. That oxadiazolone is the acid replacement the whole drug series depends on, and it is a well known bioisostere for a carboxylic acid. Compound 31j also earns its place in the claims for a reason nothing to do with the route: paragraph [0381] records that the absolute configuration of the dimethyloxane centre, assigned as S, was established by X-ray crystallography of Compound 31j. The two enantiomers had been separated earlier by chiral chromatography on a CHIRALPAK-IE column, 4.6 by 250 mm, 5 micron, running hexane and ethanol at 30 to 70 and 1 mL per minute, with the wanted isomer eluting at 9.98 minutes against 6.86 minutes for the other. 31j is the compound that settled which one was which.

Claim 4 covers Compound 31k, 5-[(4S)-2,2-dimethyloxan-4-yl]-1-[(1S,2S)-2-methyl-1-(5-oxo-4H-1,2,4-oxadiazol-3-yl)cyclopropyl]indole-2-carboxylic acid, recorded at m/z 401 in negative ion mode. This is the endpoint of the sequence and the only one of the four with a free carboxylic acid. It is the piece that actually reacts.

Claim 5 is the odd one out and the most interesting. It reads as a compound of formula 31k in accordance with claim 4 “when used in a process to make compound 31l”, the process comprising converting compound 11k to compound 11l and reacting 11l with 31k. That is a product claim limited by the use it is put to, a drafting form Australian practice still permits and which reads strangely to anyone used to European or US claim sets. Compound 31l is described at paragraph [0376] as the 2-oxoimidazole reagent used in the synthesis of Example Compounds 31 to 40, and is recorded at m/z 735.

What makes 31k valuable is how often it turns up. Besides feeding 31l, the specification uses Compound 31k obtained in Step 31-8 to make Compound 42g and Compound 48a as well, by the same amide coupling described as “an operation similar to Step 1-10 of Example 1”. Examples 54 to 73 are likewise described as made by coupling an amine derivative with a carboxylic acid derivative, and the substituent pattern recorded for Example Compound 67, the compound that became the oral GLP-1 agonist orforglipron, is precisely the pattern that Compound 31k supplies. The specification does not spell that pairing out in a dedicated step, so the connection is drawn from the names rather than stated, but the acid half of orforglipron and Compound 31k are the same fragment.

The compounds made downstream from 31k are potent. In the specification’s cyclic AMP assay against a human GLP-1 receptor cell line, Example Compounds 31 to 40, all built from Compound 31l, record half maximal effective concentrations of 0.74, 0.94, 1.6, 3.0, 3.2, 2.1, 2.5, 1.7, 1.0 and 2.2 nanomolar. Example 31 at 0.74 nM is among the three strongest in a table of eighty. None of that activity belongs to the four claimed compounds themselves.

Key Features

• Four named compounds instead of a formula. Claims 1 to 4 each draw a single structure and add only “or a salt thereof”, with no substituent variables at all. That is the opposite drafting choice to the sibling application AU 2026202208, whose claim 1 is a Markush formula with sixteen variable positions.
• An amide used as a temporary mask. The N-methyl-N-phenyl amide on compounds 31f, 31i and 31j is not part of the finished drug. It carries the carboxylic acid through the alkylation, cyclopropanation and oxadiazolone steps and is hydrolysed off at Step 31-8 to give 31k.
• A nitrile converted into an acid replacement. Compound 31i’s cyano group becomes the 5-oxo-4H-1,2,4-oxadiazol-3-yl ring of compound 31j, the acidic heterocycle that every lead compound in the family carries.
• Stereochemistry proved by crystal structure. The S configuration at the dimethyloxane centre was assigned by X-ray crystallography of Compound 31j, after the enantiomers were resolved on a CHIRALPAK-IE column with retention times of 9.98 and 6.86 minutes.
• A product claim tied to a downstream process. Claim 5 claims compound 31k when used to make compound 31l, a use limited product claim rather than a process claim, and the only claim in the application that mentions a process at all.
• No pharmacology anywhere in the claims. Nothing in claims 1 to 5 refers to GLP-1, to a receptor, to diabetes, to a composition or to a method of treatment, despite the title and abstract of the document being entirely about those things.

Who Is Behind It

The applicant is Chugai Pharmaceutical of Tokyo, majority owned by Roche since 2002 and still separately listed. The title page prints the name Chugai Saiyaku Kabushiki Kaisha, which appears to be a transcription error for Chugai Seiyaku Kabushiki Kaisha, the romanisation the company uses itself. Seventeen inventors are named, led by Hitoshi Yoshino. Wrays Pty Ltd of Perth is the Australian agent.

Paragraph [0000] sets out the whole chain, which saves any guesswork. This application is a divisional of AU 2024201884 filed 22 March 2024, which was a divisional of AU 2022205222 filed 13 July 2022, which was a divisional of AU 2020223687 filed 26 August 2020, which was a divisional of AU 2017330733, the Australian national phase entry made on 21 February 2019 from PCT/JP2017/034620. That PCT was filed 26 September 2017 and published as WO 2018/056453, claiming priority from Japanese provisional JP 2016-187605 filed 26 September 2016. Priority country is Japan.

One detail worth flagging for anyone tracking the family. The summary section of this specification sets out eight separate compound embodiments: 11j, 11k, 31f, 31i, 31j, 31k, 67a and 67b. The claims pick up only four of them, the 31 series. Compounds 11j and 11k are the pyrazolopyridine half of the molecule and 67a and 67b are the finished amine used for Example Compound 67, and they remain described but unclaimed here. Given the pattern of this family, another divisional aimed at them would be an unsurprising next filing.

Why It Matters

Intermediate claims rarely make headlines but they do a lot of work. A compound claim on a finished drug stops someone selling that drug; it does not stop a contract manufacturer in a third country making and exporting the acid fragment. Claims on 31f, 31i, 31j and 31k reach that activity directly. They also reach anyone who develops a competing molecule that happens to use the same indole fragment, which matters here because the oxadiazolone-substituted cyclopropylindole is a distinctive piece of medicinal chemistry rather than a generic building block.

The timing is the other half of the story. The priority date is September 2016 and the family expires from the 2017 international filing, so these claims have less than a decade to run. Filing them in 2026, with orforglipron approved and marketed, looks less like protecting a research programme and more like tidying the perimeter of a product before generic manufacturers start planning routes. The synthetic detail in the specification, down to reagent volumes and 93 per cent isolated yields, is exactly what a generic chemist would work from, and it has been public since 2018.

It is also worth being blunt about what this document is not. Nothing in it is new relative to the 2017 disclosure. The examples, figures, assay tables and monkey data are identical to those in the parent applications and in the sibling AU 2026202208. The only thing that distinguishes the two applications published on 9 April 2026 is the set of claims drawn around a shared description: 2026202208 takes the formula and the therapeutic uses, 2026202206 takes the pieces the drug is built from.

Related Concepts

• Reaction intermediate – the class of compound all five claims are directed at, valued for blocking manufacturing routes.
• Indole – the bicyclic ring system common to compounds 31f, 31i, 31j and 31k.
• Oxadiazole – the ring family containing the 5-oxo-4H-1,2,4-oxadiazol-3-yl group introduced at compound 31j.
• Orforglipron – the oral GLP-1 agonist whose acid half corresponds to claimed compound 31k.
• X-ray crystallography – the method used on compound 31j to assign the S configuration of the dimethyloxane centre.
• Divisional patent application – the filing route that produced this application and its sibling from a shared 2017 description.

---

AU 2026202206 was published in the Australian Official Journal of Patents on 9 April 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.