Predicting GLP-1 Response: building the foundation for precision prescribing

GLP-1 drugs are among the fastest-growing drug classes globally, yet more than 50% of patients discontinue within 12 months. When patients stop therapy, many rapidly lose the metabolic and cardiovascular benefits they gained.

In the US alone, approximately $72 billion was spent last year on GLP-1 therapies. High discontinuation and variable efficacy mean a substantial proportion of that spend does not translate into sustained health benefit.

A predictive test could help ensure the right patients receive the right therapy that they will find safe, efficacious and tolerable, improving outcomes while reducing clinical and financial waste.

The Phase 1 POC analyzed whole genome and limited clinical data from 4,600 GLP-1 RA patients and demonstrated that we can:

• Identify strong and weak responders based on genetic signatures
• Quantitatively predict efficacy using BMI and HbA1c change
• Detect over 2,500 genetic signatures mapping to 1,100 genes
• Identify 15 major biological mechanisms driving response

This is the first time detailed mechanistic efficacy signals have been shown directly within a GLP-1-treated real-world population at this level of resolution.

Most existing tests rely on small gene panels derived from GWAS or known appetite-related genes.

This collaboration analyzes quantitative drug response directly in GLP-1-treated patients, identifying a large number of combinatorial genetic patterns across multiple biological pathways and tissues.

Rather than testing a handful of well-known genes, the platform uncovers much deeper mechanism-level drivers of response and non-response, including multiple novel pathways beyond classical GLP-1 biology. As well as efficacy for weight loss, glycemic control and other phenotypes, we are also examining the safety and tolerability of these drugs.

We can also use this to develop drug specific tests, looking at the relative performance and benefits of one compound over another for specific patients.

The many pathways identified include:

• Dopaminergic signaling
• Glutamatergic signaling
• Leptin-mediated pathways
• PI3K-AKT signaling
• Histamine signaling
• Insulin secretion and lipid metabolism

These mechanisms influence appetite regulation, reward processing, energy balance, glucose homeostasis, and metabolic control. Importantly, some identified genes fall outside established GLP-1 mechanisms, suggesting novel biology.

By identifying biological mechanisms rather than isolated genes, the analysis helps explain why some patients experience strong, durable benefit while others do not, supports quantitative prediction of response magnitude and tolerability, and provides a mechanistic foundation for more precise prescribing and drug-specific stratification as the GLP-1 class continues to expand.

Phase 2 expands the dataset to up to 25,000 patients and bring in more detailed longitudinal clinical data.

It will:

• Validate the efficacy signals at larger scale, and at higher resolution
• Identify predictive markers of safety and tolerability
• Assess drug-specific differences in different applications
• Evaluate discontinuation and rebound risk
• Enable head-to-head drug comparisons as data matures

This moves the program from proof-of-concept to clinically actionable stratification.

Payors face significant losses associated with GLP-1 coverage. Large US insurers have reported hundreds of millions in annual losses.

A predictive test could:

• Reduce ineffective prescriptions and improve the affordability of healthcare
• Lower tolerability issues (side-effects) and discontinuation rates
• Support sustainable expansion of coverage to patients who will benefit long-term
• Enable differentiated reimbursement strategies for their health plans

This shifts GLP-1 prescribing from trial-and-error to evidence-guided allocation.

Biopharma companies face:

• Massive competition and potential market saturation with over 500 GLP-1-related clinical trials across 100+ indications
• Downward price pressure
• Highly restricted reimbursement and resistance to new indications

Mechanism-based stratification can:

• Enrich clinical trials for responders (response biomarkers)
• Identify strong responder subgroups (complementary diagnostic)
• Differentiate competing assets
• Support development in new indications (biomarker-led indication switching)
• De-risk late-stage programs (clinical trial design)

Past efforts to expand GLP-1s into new indications - such as the semaglutide Alzheimer’s trial - largely targeted broad patient groups defined by their clinical symptoms or diagnostic labels. These approaches do not account for the fact that complex chronic diseases are driven by multiple distinct biological mechanisms, and that patients have different genetic and comorbidity backgrounds meaning they will respond differently to modulation of GLP-1 pathways.

PrecisionLife’s platform finds more pathways and applies mechanistic patient stratification, identifying the specific combinations of genetic and biological drivers that create different subgroups within a disease. This allows us to compare the broader range of biology to find true candidates for GLP-1 approaches.

This reveals which subpopulations in new indications are most likely to be responsive to GLP-1s and which are not, and which have the highest likelihood of specific adverse events. By matching therapies to the biological mechanisms active in each subgroup - rather than treating all patients as a homogeneous population - PrecisionLife enables far more targeted, de-risked development in secondary indications.