The discovery of lithium remains one of the most intriguing stories of modern and not so modern medicine. As this volume illustrates, it has important lessons for psychiatry at several levels. This final chapter summarizes snapshots of the key issues for the future.

Lithium illustrates the phenomenon of serendipity in the discovery of novel treatments. This is important because we face an immediate future where major pharmaceutical companies have, with only a few notable exceptions, withdrawn from research and development in psychiatric disorders. That means that there could well be nothing that is really new for a generation. It reflects a failure of the preceding two decades of brain research to generate the concrete scientific understanding that best underpins applied research in our field.

In the future we will continue to need what I like to call ‘guided serendipity’. In other words, discovery that is fortuitous, but nevertheless guided by approximate hypotheses that are plausible and science led. The unexpected use of ketamine to treat depression is the most recent really interesting example. Ketamine is a drug with an established indication that is being ‘repurposed’, to use the jargon buzzword. There will continue to be innovations in neuroscience, targeted mainly at neurodegeneration. This may be primarily as ‘biologicals’, such as antibodies, aptamers, RNA antagomirs and cell therapy. However, for all we know, these approaches and the development of more familiar types of drug may offer treatments that turn out to work in psychiatric indications. Guided serendipity, having served us well in the past, may yet do so again.

For this to happen, however, we must be free to experiment in the spirit of John Cade. Much modern regulation throttles experiment and hence innovation. It does so by demanding standards of safety or certainty that are impossible, or impossibly expensive, to satisfy. Too often as a Principal Investigator, I have found myself engaged in the pointless exercise of being required to guarantee to prevent from happening something that has never happened before or to collect detailed data on things that we know have happened before and that are bound to happen again, such as adverse effects.

Quite apart from the generic regulatory barriers to conducting experimental medicine, bipolar disorder presents a particular additional challenge because it is defined to be an episodic disorder. Studying relatively infrequent severe episodes is very challenging. It means either trials in acutely ill patients or relapse-prevention studies that require years rather than weeks of follow-up. There is thus currently no accepted way to do small-scale, short-term, proof-of-concept studies in bipolar patients, and I would argue considerable pessimism about innovation primarily for this reason. This is not a new situation of course. It partly explains why the risk of developing a drug for a bipolar indication is usually only undertaken by industry for products (e.g. antipsychotics and anticonvulsants) that already have indications.