Merck scientists, with quantitative systems pharmacology support from Rosa & Co., have published a new paper in Pharmaceutics titled “In Silico Hypothesis Testing in Drug Discovery: Using Quantitative Systems Pharmacology Modeling to Evaluate the Therapeutic Value of Proinsulin Conversion to Insulin Therapy for Type 2 Diabetes Mellitus.”
The work reflects Rosa’s use of transparent, fit for purpose QSP modeling to answer early go or no go questions in metabolic portfolios.
Key findings
-
The team extended a published QSP model of glucose homeostasis to include proinsulin biology, created virtual patients representing key type 2 diabetes phenotypes, and ran virtual clinical trials of a hypothetical therapy that converts circulating proinsulin to insulin.
-
Across virtual patients and disease stages, the mechanism produced only about a 0.2 percent reduction in HbA1c, below commonly accepted thresholds for meaningful clinical benefit.
-
To reach a 0.5 percent or greater HbA1c improvement, the model required proinsulin to insulin ratios that exceed reported physiological ranges, suggesting the mechanism is unlikely to justify a drug development program.
Why this matters for drug developers
The study highlights how QSP based in silico hypothesis testing can quantify the ceiling of clinical benefit for biologically attractive mechanisms, show when biomarker changes are unlikely to translate into meaningful outcomes, and help T2D and obesity teams focus discovery and BD resources on concepts with a realistic path to differentiation.
Read the full paper
The article is open access and provides a detailed example of how virtual patients and QSP models can be used to test “what if” mechanisms in type 2 diabetes before major investment in discovery and clinical studies.
Read the article on the publisher’s site: “In Silico Hypothesis Testing in Drug Discovery: Using Quantitative Systems Pharmacology Modeling to Evaluate the Therapeutic Value of Proinsulin Conversion to Insulin Therapy for Type 2 Diabetes Mellitus,” Pharmaceutics 2025, 17(12), 1522.