Prescription drugs cost more in the United States than anywhere else in the world. In 2021, the median launch price for a new drug was $180,000 for a year’s supply. In 2022, in an effort to reduce the price paid by patients and payers, Congress included provisions in the Inflation Reduction Act (IRA) that allow Medicare to negotiate drug pricing. In August 2023, the Centers for Medicare & Medicaid Services announced the first ten drugs selected for negotiation, which collectively accounted for about $50 billion in Medicare Part D coverage from June 2022 to May 2023.
Several large pharmaceutical companies are challenging these provisions in court, and it is not yet clear if the provisions will stand or be struck down. However, it is clear that the provisions in the IRA, as well as California’s effort to develop its own, low-cost insulin biosimilar, are the first steps in a longer effort to reduce prescription drug costs.
Drug prices are high in the U.S. for a variety of reasons, but a key factor is that approved drugs must generate sufficient revenue to cover not only their cost of development, but the costs of failed development efforts as well. Only about 12% of drugs entering clinical trials ultimately receive FDA approval, with a typical development time frame of 10+ years and cost of $1B+ billion. A recent analysis published in JAMA estimated that $50-$60 billion dollars is spent annually on failed oncology clinical trials alone. Consequently, the pharmaceutical industry argues that the reduction in revenue resulting from price control mechanisms such as those in the IRA will limit their ability to invest in R&D, especially in the small-market and rare disease indications that typically have the greatest unmet medical need.
In litigating the cost containment provisions of the IRA, the pharmaceutical industry and the U.S. government are fighting over the latch on the barn door as the horses prepare to charge. Substantially reducing the price of prescription drugs requires novel R&D and clinical trial paradigms that actually reduce the sunk cost of drug development.
One promising solution to this pressing issue is the new era of compute-enabled biotechnology companies that integrate machine learning (ML), deep neural networks (DNN), and artificial intelligence (AI) to not only predict the safety and efficacy of new drugs, but discover, design, and develop better therapies for patients. This new breed of fullstack, compute-enabled biotechnology company can utilize AI across the entire drug discovery and development process, spanning drug target/lead optimization and biosimulation modeling for toxicity testing to patient stratification and clinical trial enrollment. This can significantly reduce the time, cost, and risk of drug development and enable earlier and more informed go/no-go decisions.
The dawn of the compute-enabled biotech company has created a new era of programmable biology in which we can utilize computational methods in conjunction with disciplines such as synthetic biology to radically transform the cost, pace, and output of the drug development process. To date, computational biology methods have been based on curated inputs — i.e. researchers select large data sets that are analyzed using algorithms to detect pre-specified outputs. While this has substantially simplified the analysis of large and complex data sets and enabled the discovery of novel disease mechanisms and targets for therapeutic development, the approach is still limited by the selected data inputs and desired outputs.
More recently, generative AI-based approaches (similar to those used in ChatGPT) have enabled the de novo identification of important biologic pathways, drug targets, and the formulations/structures of new therapeutic molecules. These approaches aren’t restricted by particular data inputs or specified outputs, and are designed to provide insights based on disparate pieces of data gleaned from diverse sources, such as published literature, scientific databases, patient registries, etc.
Other companies are integrating patient-derived clinical data and biospecimen samples with proprietary genome sequencing, AI, and synthetic biology technologies to discover novel disease biomarkers and enable the development of novel therapies. These approaches radically shorten development time and reduce development costs and risks, while also enabling wholly new classes of drugs with the potential for substantially improved efficacy and safety.
AI/ML technologies are also being used to select patients based on their specific health and disease profile. These approaches analyze millions of health records and published data sets to identify demographic, disease, and treatment-related information to identify patients most likely to have optimal responses to an investigational therapy. Such highly targeted patient selection can reduce the size, scope, duration — and, ultimately, the cost — of clinical trials by increasing the likelihood of positive outcomes.
While legacy pharma companies battle in court with government agencies over how to address the costs that result from antiquated drug development paradigms, a growing cadre of compute-enabled life science companies are unlocking the nascent power of next-generation compute technologies to transform drug discovery and development and creating a new era of accelerated R&D that can ultimately bring more cost-efficient, efficacious therapies to market for millions of patients. It is these tech-enabled life sciences companies that are creating a future that all of us —patients, payoers, industry, and investors —need for long-term health.
Photo: metamorworks, Getty Images