ASP4345

ASP4345 is a small molecule from Astellas, built to treat the cognitive impairment that comes with schizophrenia. Antipsychotics control hallucinations and delusions but do little for attention, memory and planning, and no drug is approved for those deficits [3]. The dopamine D1 receptor in the prefrontal cortex has been a leading target for decades [3][6]. Drugs that activate it directly were held back by poor absorption, low blood pressure and seizures [3][6]. ASP4345 takes the other route. It is a positive allosteric modulator, a compound that binds somewhere other than dopamine's own site and makes the receptor respond more strongly to the dopamine already being released [2][6].

The published preclinical record is thin. Astellas's patent reports that the compound doubled dopamine's potency at the human D1 receptor at 0.46 µM, and that 0.1 mg/kg by mouth was effective in mice whose short-term memory had been disrupted with MK-801 [1]. The phase 1 paper mentions other rodent models of cognitive deficit only as data on file [2]. No primary pharmacology paper has followed.

Phase 1 went well. Single doses of 3–900 mg in healthy volunteers peaked in the blood within about two hours, cleared with a half-life of 9–14 hours, and reached the cerebrospinal fluid [3]. A second study gave 3, 15, 50 or 150 mg a day for 14 days to 36 patients with schizophrenia or schizoaffective disorder who stayed on their antipsychotics, with 12 more on placebo [2]. Adverse events were about as common on placebo as on the drug. On the Cogstate battery, patients on every dose improved more than placebo on the Detection task, a response-speed test of psychomotor function, with effect sizes of 0.58–0.99 on day 14 that fell to 0.14–0.34 three days after the last dose [2]. A smaller effect appeared on the Identification task, a test of visual attention. The One Back test of working memory showed no consistent benefit, which the authors called surprising given dopamine's role in working memory [2]. EEG markers of auditory processing moved in the hoped-for direction at three of the four doses, without reaching statistical significance [2]. The authors were explicit that the study was sized for safety and that no conclusion about efficacy could be drawn from it [2].

Phase 2 found nothing. The trial randomised 233 clinically stable patients at 27 sites in the United States to 50 mg, 150 mg or placebo once a day for 12 weeks on top of their antipsychotics [4]. The primary measure was the MATRICS Consensus Cognitive Battery composite, on which patients started about 1.7 standard deviations below the healthy average. It rose by 1.15 points on placebo, 1.34 on 50 mg and 0.87 on 150 mg. The differences from placebo were 0.18 and −0.28 points, with p values of 0.86 and 0.78 [4]. The secondary measure, a performance test of everyday skills called the UPSA-2-ER, showed the same absence of effect [4]. Astellas stopped development [5][7].

Why it failed is not known. No analysis of the phase 2 data has been published. A review of the D1 target that still described the trial as under way had noted the main weakness of the approach: a modulator can only amplify dopamine that is released, and imaging studies find dopamine release blunted across the cortex in schizophrenia [6]. The same review argued that antipsychotic treatment brings down the raised cortical D1 receptor levels found in untreated patients, that D1 levels fall with age, and that early illness may therefore offer a better window than chronic illness [6]. The phase 2 patients averaged 43 years of age and were all on antipsychotics [4]. The direct D1 agonists tested in patients with schizophrenia, DAR-0100A and PF-06412562, had not separated from placebo on their main cognitive measures either [6].

It amplifies dopamine and does not replace it. In the patent's assay, CHO cells carrying the human D1 receptor were given a range of dopamine concentrations with and without test compound, and cyclic AMP was measured. The compounds moved dopamine's concentration–response curve to the left, more so at higher concentrations, and had no agonist effect when dopamine was absent [1]. Potency was expressed as the concentration that doubled dopamine's potency, which was 0.46 µM for the hydrochloride of ASP4345 [1]. That looks modest beside other D1 modulators, though the assays differ: a 2019 review lists Lilly's DETQ at 5.8 nM and a Bristol-Myers Squibb compound at 43 nM, and calls the 460 nM Astellas compound moderately potent [8].

Why the D1 receptor. D1 receptors sit on the dendritic spines of the glutamate neurons of the prefrontal cortex, and working memory depends on how strongly they are stimulated. The relationship is an inverted U: too little stimulation leaves the neurons that hold information online barely firing, and too much, as under stress, suppresses them [6]. That narrow window is one argument for a modulator. Another is selectivity, since the site where dopamine binds is closely conserved across dopamine receptors and allosteric sites are less so [3]. A third is timing: a modulator keeps the pattern of dopamine release that behaviour produces, where an agonist stimulates the receptor continuously [6]. The authors of the pharmacokinetic paper compare it to a dimmer switch [3].

Where it binds is unknown. The phase 1 paper says so directly [2]. Other D1 modulators, including DETQ and the Bristol-Myers Squibb compound, are thought to bind a cleft in the receptor's second intracellular loop, and structural work points to two possible allosteric sites [8][9]. The published papers describe ASP4345 as selective for D1 without showing binding or functional data at other receptors [2][3].

The mouse result leaves questions open. Several D1 modulators work poorly on rodent receptors. DETQ is about 30 times weaker at the rat and mouse receptor than at the human one and is inactive in ordinary mice, and the Bristol-Myers Squibb compound lacks rodent activity altogether [8]. The Astellas patent reports a behavioural effect in mice but measured potency only at the human receptor, and its maze experiment has no D1-blocking control and gives no group data, only the effective dose [1].

In patients the brain signals were suggestive and unconfirmed. The 14-day trial recorded EEG measures of early auditory processing. The 40 Hz auditory steady-state response, a measure of how well cortical gamma rhythms lock to a click train, showed a growing effect size against placebo by day 14 at 3, 15 and 150 mg. Mismatch negativity, the brain's automatic response to an odd tone, showed a small to moderate effect by day 7 at the same three doses [2]. The 50 mg dose improved neither, the P3a response did not change at any dose, and the study was too small for any of these to reach statistical significance [2].

It did not move blood pressure. Direct D1 agonists lower blood pressure, because D1 receptors relax renal and other blood vessels [6][7]. The D1 modulator mevidalen did the opposite, raising blood pressure and heart rate at 75 mg and above [7]. ASP4345 produced no clinically relevant change in vital signs at 3–150 mg [2][7].

The compound that went into people is the monohydrochloride salt [2][3]. It is the first worked example in Astellas's patent, made by coupling (5-chloro-2-oxo-1,3-benzoxazol-3-yl)acetic acid to the benzimidazole methylamine and treating the product with hydrogen chloride, and one claim covers that salt specifically [1].

Trial capsules were hard gelatin, in strengths of 0.5, 5 and 50 mg, with anhydrous silicic acid and low-substituted hydroxypropyl cellulose as excipients [3]. Exposure rose less than in proportion to dose from 3 to 900 mg, and the 900 mg dose was the most variable, which the authors suggest may reflect the number of capsules that dose required [3]. A high-fat breakfast delayed absorption of a 300 mg dose without changing how much was absorbed [3].

Tolerability was close to placebo in every published dataset. In the 14-day trial 29 of 36 patients on ASP4345 reported an adverse event, against 9 of 12 on placebo, and the rate did not rise with dose [2]. Headache was the commonest, in 9 of 36 on the drug and 4 of 12 on placebo. Somnolence was reported by 7 of 36 and 1 of 12, and constipation by 5 of 36 and none on placebo [2]. Nearly all events were mild. One patient on 3 mg had a moderate psychotic episode, and one on 50 mg had a small-bowel obstruction five days after the last dose that was judged unrelated [2].

Over 12 weeks, 28 of 65 patients on 50 mg and 28 of 68 on 150 mg reported an adverse event, against 45 of 100 on placebo [4]. Only two events were common enough to be listed. Headache was evenly spread, and dizziness was not: it was reported by 4 of 68 on 150 mg, 1 of 65 on 50 mg and nobody on placebo [4]. Five serious events occurred, none judged drug-related: a head injury, an acute respiratory failure and two psychotic episodes on ASP4345, and one case of suicidal ideation on placebo [4]. Eleven patients stopped because of an adverse event, six on 50 mg, one on 150 mg and four on placebo [4]. Nobody died in either patient trial [2][4].

The problems of D1 agonists did not appear. Earlier D1 agonists were limited by seizures and falls in blood pressure [3]. Neither is reported for ASP4345, and vital signs, ECGs and laboratory tests showed no clinically relevant change over 14 days [2]. Movement-disorder ratings did not worsen in either trial, symptom scores did not change in the 14-day trial, and neither trial found a signal on the suicidality scale [2][4]. A checklist used to screen for abuse potential showed no drug-related change [2].

What is missing. Exposure stops at 12 weeks. Every patient was also taking an antipsychotic, so nothing is known about ASP4345 given alone for more than a single dose [2][3][4]. Animal toxicology is described only as "well tolerated" [2].