Biotechnology and Pharmaceuticals · global
AI-Designed Oral Molecule Targets Lipoprotein(a) as ISM0900 Enters Preclinical Development
Insilico Medicine says ISM0900 reduced lipoprotein(a) by up to 83.9% in preclinical studies and demonstrated oral pharmacokinetics. The candidate offers another possibility in a field dominated by injectable therapies, but all data still await independent review and validation in human studies.
Lipoprotein(a), abbreviated as Lp(a), is a cardiovascular risk factor that is not easily modified through diet or exercise. Insilico Medicine has now selected the AI-assisted small molecule ISM0900 as a preclinical candidate, aiming to turn Lp(a)-lowering treatment into an oral drug. If it can enter human trials and maintain its effect, dosing convenience will be key to competing with most nucleic acid-based injectable therapies.
Lp(a) is a lipoprotein particle containing apolipoprotein(a), and its concentration is primarily influenced by genetics. Elevated Lp(a) is associated with increased risks of atherosclerotic cardiovascular disease, ischemic stroke, and aortic valve stenosis. However, whether lowering Lp(a) levels necessarily translates into fewer heart attacks or strokes still needs to be answered by large clinical endpoint trials.
According to data released by the company, ISM0900 was designed and optimized across multiple properties with the involvement of a generative AI platform. It demonstrated nanomolar-level activity in vitro and oral pharmacokinetics. The company also said the molecule reduced Lp(a) by up to 83.9% in preclinical studies while showing a wider predicted safety margin. However, the announcement did not fully disclose the number of animals, doses, statistical analyses, or complete toxicology results for the individual experiments.
Candidate nomination means the team has selected, from among multiple lead compounds, a molecule to proceed into formal preclinical development. It does not mean the molecule has been approved to enter human trials. ISM0900 must still complete work including manufacturing process development, repeat-dose toxicology, safety pharmacology, and dose estimation before an application for human trials can be submitted. No participant safety or efficacy data are currently available.
Lp(a) drug development in recent years has focused mainly on antisense oligonucleotides and small interfering RNA. These therapies can substantially lower Lp(a) in humans but generally require subcutaneous injection. If an oral small molecule can balance the magnitude of reduction, selectivity, and long-term safety, it may reduce the treatment burden. On the other hand, small molecules are more likely to interact with multiple proteins. Whether ISM0900 interferes with systems such as plasminogen, which is structurally similar to apolipoprotein(a), will be an important question in its safety assessment.
ISM0900’s 83.9% reduction should therefore be regarded only as an early development signal and cannot be directly compared with human trials using different models, doses, or observation periods. The announcement was also not supported by independent research on the same development or by a peer-reviewed paper. Its real test will begin with the first human trial: whether oral exposure reaches the expected level, whether the reduction in Lp(a) can be reproduced, and whether long-term use is safe will determine whether this AI-designed molecule can move from impressive preclinical numbers into cardiovascular treatment.