When contemplating the contemporary successes of biological psychiatry, it is easy to forget that psychiatric biomarker discovery has an almost century-long history. Scant attention has been paid to these historical insights, in contrast to the deferential attention that has been paid to early attempts at nosology, such as those of Kraepelin. What lessons can we glean from this forgotten history?

In one of the first documented studies, Segal and Hinsie in 1926^{Reference Segal and Hinsie1} noted that levels of oxygen in the blood of people with psychosis were reduced. Another early biomarker reported in 1932 in psychosis was acidosis indexed by lowered pH.^{Reference Hurst2} A lowered metabolic rate and reduced oxygen consumption in psychosis/dementia praecox was also reported in 1932.^{Reference Stevenson and Thomas3} In 1934, Looney and Childs^{Reference Looney and Childs4} described elevated lactate and reduced glutathione in people with schizophrenia. Lowered basal metabolic rate was again reported in 1937.^{Reference Thompson, Corwin and Aste-Salazar5} While the choice of these markers was clearly predicated on what was available at the time rather than a priori hypotheses, they converge on abnormal energy generation. It is probably accurate to say that these early research findings were also methodologically limited by contemporary standards and have had minimal or no enduring impact on contemporary psychiatric research. However, is this a fair reflection of the importance and promise of these findings?

The focus on metabolic abnormalities in schizophrenia is nonetheless echoed by one of the earliest biological therapies, insulin shock^{Reference Larkin6} therapy pioneered by Manfred Sakel. It needs to be noted that insulin coma therapy originated after observations of mental clarity in people who had emerged from coma rather than arising from these biological markers, and its conduct was not a highpoint in psychiatric ethics or practice. Nevertheless, insulin resistance is now understood as part of the pathophysiology of many major psychiatric disorders.

These ideas fell off the radar. Seymour Kety, past head of the National Institute of Mental Health (NIMH), criticised bioenergetic failure in an article in Science back in 1959, writing: ‘it is difficult for some to believe that a generalized defect in energy metabolism - a process so fundamental to every cell in the body - could be responsible for the highly specialized features of schizophrenia’.^{Reference Kety7} We realise now that Kety did not consider that the brain is unique with its high energy demands, limited energy reserve and vulnerability to oxidative stress. He also did not consider that mitochondrial disorders can affect specific organ systems, for example, Leber hereditary optic neuropathy, and that specific neuronal structures could be affected in a similar way or the bidirectional interaction of bioenergetic dysfunction with other operative pathways, such as inflammation, oxidative stress, neurogenesis or apoptosis.

These ‘archaeological’ findings presaged a renaissance in metabolic psychiatry almost a century later. There are neuroimaging data using methodologies such as F-18-deoxyglucose and positron emission tomography showing decreased glucose metabolism and oxygen consumption in many major psychiatric disorders, such as schizophrenia, indicative of a hypometabolic state,^{Reference Townsend, Pillinger, Selvaggi, Veronese, Turkheimer and Howes8} both widespread and in specific brain regions such as the prefrontal cortex. The finding of reduced glutathione anticipated subsequent discoveries of oxidative stress and reduced oxidative defence is in major psychiatric disorders. Equally, the finding of increased lactate has been validated by meta-analysis of contemporary neuroimaging studies. In parallel, there is meta-analytic evidence of an acidotic state and lower brain pH in several^{Reference Hagihara, Murano and Miyakawa9} psychiatric disorders. Taken together, this evidence reiterates that many major psychiatric disorders, including schizophrenia, bipolar disorder and autism, share abnormalities in mitochondrial energy generation. This is characterised by a shift from more efficient aerobic to less efficient glycolytic energy generation – the Warburg effect.

These findings have clinical echoes. It is now established that psychiatric disorders are associated with metabolic dysfunction, which clinically manifests as the metabolic syndrome, an amalgam of several clinical features including abdominal obesity and a body mass index (BMI) above 25, high triglycerides and low high-density lipoprotein (HDL) cholesterol, hypertension, elevated fasting blood glucose and insulin resistance, as well as fatty liver. While the initial focus was on these features as being adverse effects of antipsychotic agents, we now know that these features are more common in people with first break illness and in the prodrome, suggesting that they form part of the wider disease phenotype.

We now know that many psychiatric disorders from schizophrenia to bipolar disorder and autism share mitochondrial dysfunction. Mitochondria play an essential role in energy production, oxidative stress regulation and apoptosis, which are crucial for brain function. Evidence of mitochondrial dysfunction derives from mitochondrial DNA (mtDNA) abnormalities such as higher rates of mtDNA mutations or deletions and an increased burden of heteroplasmic mtDNA variants, which can affect mitochondrial function. Energy metabolism deficits are indicated by lowered adenosine triphosphate levels, especially in brain regions involved in mood and cognition, such as the prefrontal cortex. Mitochondria are the primary source of reactive oxygen species, which lead to oxidative stress, which is also found in psychiatric disorders such as depression, schizophrenia and autism. Post-mortem brain studies, especially in schizophrenia and bipolar disorder, show abnormalities in mitochondrial structure and reduced density of mitochondria in neurons. Increased expression of proteins related to mitochondrial fission or fusion have been found in people with schizophrenia, major depressive disorder and bipolar disorder. Certain medications, such as lithium, are believed to exert part of their therapeutic effects by improving mitochondrial function. Mitochondrial dysfunction can trigger the release of pro-inflammatory cytokines, contributing to neuroinflammation, which is often observed in disorders such as depression and schizophrenia, and is similarly linked to oxidative stress, epigenetic changes and telomere shortening.

Almost a century after the discovery of the biomarkers of energy generation, a new generation of studies has been commenced. These now target mitochondrial energy generation, prompted by these contemporary biomarker findings. An exemplar is the ketogenic diet, which is attracting significant popular and early research attention. Medications targeting mitochondrial function, such as antioxidants (e.g. N-acetylcysteine, coenzyme Q10) and metabolic modulators (e.g. creatine, carnitine), have been explored as treatments for psychiatric disorders. Theoretical bottom-up stem cell-based discovery platforms have also detected mitochondrial agents such as trimetazidine as potential therapeutics, which may reverse the Warburg effect. Preliminary evidence suggests some of these might reduce symptoms, further supporting the role of mitochondrial dysfunction.

In summary, the earlier biomarker discoveries were both visionary and neglected. These early pioneers merit a re-evaluation of the accuracy of their contributions, the prescience of their discoveries and the clinical translational impact of this knowledge.