Discovering a new drug is often cited as one of the most challenging feats of engineering ever conceived. The reason lies in the extraordinary complexity of the human body. It’s an intricate network of specialised organs, each with unique functions, filled with specialised cells that operate like miniature cities. Within each cell, there are networks and feedback loops, every role performed by a different protein – a molecular machine in its own right.

These proteins don’t merely perform static tasks; they engage in dynamic interactions governed by electrodynamics, fine-tuned actions, and communication channels that can only exist within the delicate balance of cellular environments. The level of precision required to interact with these processes – to alter just one protein’s behaviour without disrupting countless others - is almost unimaginably high.

Diseases, particularly complex ones like cancer, add another layer of difficulty. They wield one of nature’s most potent weapons: evolution. Over generations, cancers evolve, “testing” thousands of genetic mutations until the right set produces growth, survival, or resistance traits that give them an advantage. It’s a trial-and-error process, ruthlessly effective over time. Against this biological force, drug discovery is an attempt to interrupt or manipulate highly evolved systems – many of which have adapted specifically to evade our best efforts.

In early drug development altering a molecule’s solubility, for example, can affect its absorption, while increasing binding affinity may introduce unforeseen interactions elsewhere in the body. Every step in this complex chain requires meticulous tuning, as even the smallest change can cascade into unintended consequences, making the journey from lab to treatment an intense and uncertain process. In late drug development, the complexity of the patient's disease biology can throw up unforeseen, unforeseeable challenges.

The question posed: Why do drugs fail? My response: given the complexity, how has anything ever been approved?! Drug discovery is a game of assumptions and models of models, and it is hoped that everything will translate at a certain point. Drugs fail at all points in this journey when there is a mismatch in those things.

For small biotechs and early discovery teams, drug development is often derailed by a series of obstacles collectively known as the “valley of death”. This framework – around for decades – describes the formidable gap between laboratory discovery and clinical application where countless promising compounds fall short of making it to market, including 56% of which fail due to safety failures.

This “valley” captures both scientific and financial hurdles, with companies frequently caught in a cycle of limited funding and technical setbacks. To complicate matters, structural challenges such as misaligned incentives, investor pressures, and regulatory limitations further entrench these scientific bottlenecks. And once a molecule’s chemical structure is locked in, options for adapting its formulation, dose, or other characteristics are limited, making it even harder to address any emerging issues in the later stages of development.