How Were Psychiatric Drugs Discovered? Lithium, Chlorpromazine and Imipramine, 1949–1976

This is Part 5 of the series A History of Psychiatry. Part 4 covered four treatments that acted hard on the body: fever, coma, convulsions and surgery. This part is about the three drugs that came next. Lithium, chlorpromazine, imipramine. All three are still in use, and none was made because someone had worked out the cause of a mental illness. Each was found by a person looking for something else who noticed what was happening in front of him. And unlike the treatments in Part 4, these were then put through trials with a comparison group.

This series is a history of psychiatry, not advice on diagnosis or treatment. If you are struggling with your mental or physical health, please contact a doctor or a support service. If you have concerns about a medicine you are taking, do not stop it or change the dose on your own; talk to your doctor or pharmacist. The text includes diagnoses and expressions that are no longer used. They are given as they appear in the sources of the time.

Drugs that quietened, none that treated

Until the end of the 1940s, the drugs used in psychiatric hospitals were ones that put agitated patients to sleep or dulled them. Thomas Ban, a historian of psychiatric drugs, wrote in 2007 that before the 1950s there was no effective drug therapy for mental illness.

Patients with mania (a state of abnormally raised mood, in which a person may go without sleep and stay constantly on the move) were usually given repeated electroconvulsive therapy and heavy doses of sleeping drugs. As Part 4 showed, the hospitals were overflowing. Insulin coma and lobotomy spread partly because there was nothing else to try.

Then three drugs appeared in under ten years.

Lithium: the guinea pigs went quiet (1949)

The first came from Australia. The psychiatrist John Cade suspected that the urine of patients with mania carried some toxic substance. Working alone, with a disused hospital kitchen as his laboratory, he ran experiments on guinea pigs. He wanted to know whether uric acid, one component of urine, made the toxic effect of another, urea, any worse. But uric acid barely dissolves in water. So he chose its most soluble form: lithium urate, in which it is combined with the metallic element lithium.

The toxicity turned out far lower than expected. It looked as if the lithium was protecting the animals. To check, he switched to lithium carbonate, which contains no uric acid. Urea alone killed five guinea pigs out of ten. With lithium carbonate added, all ten survived. And when he injected lithium carbonate by itself, the animals, though fully conscious, became extremely lethargic about two hours later and stopped responding to stimuli. After an hour or two they were back to normal.

Cade first took lithium himself to check that it did him no harm, then gave it to ten patients with mania. The first was a man of 51 who had been in a state of manic excitement for five years and was regarded as the most troublesome patient on his ward. He started on 29 March 1948. Within three weeks he was well enough to move to the convalescent ward. He left hospital in July and went back to his old job. Later he stopped taking the medicine, the mania returned, and he was readmitted the following January. Back on lithium, he settled again in a fortnight.

The 1949 paper sets out the course of each of the ten. Lithium was also given to six patients with schizophrenia (dementia praecox, in the language of the time) and three with depression. In the schizophrenia group, three who had been restless became quiet, but the illness itself did not improve. The three with depression got neither better nor worse. It was mania that responded.

There was no comparison group, and Cade knew it. The ideal, he wrote in the paper, was to compare enough treated and untreated patients. But mania was not that common, and a single doctor even in a large hospital would need years to gather the numbers. He also conceded that it might seem a long way from lethargy in guinea pigs to excitement in patients.

The drug was also dangerous. Cade's own paper warns that the symptoms of overdose can end in death if ignored, and that patients on the highest doses should be seen every day. As the neurologist Andrew Lees put it in a 2020 account, the dose that worked was close to the dose that poisoned. One of Cade's own patients died, and he temporarily banned lithium at his hospital. At about the same time, the American press was reporting poisonings in heart patients who had taken too much of a lithium salt sold as a substitute for table salt. Lithium treatment, Lees writes, was on the brink of being forgotten.

Chlorpromazine: an aid to anaesthesia reaches psychiatry (1952)

The second drug came from the operating theatre. Henri Laborit, a surgeon in the French army, was looking for a way to prevent surgical shock, the sudden collapse that can follow an operation. His approach was to add to the anaesthetic drugs that lowered body temperature and damped the body's reactions.

On 11 December 1951, Paul Charpentier synthesised chlorpromazine at the French drug company Rhône-Poulenc. It came out of work on antihistamines (allergy drugs) and was expected to strengthen the effect of anaesthetics. When Laborit injected it into patients, he noticed something. They did not lose consciousness and were only slightly sleepy, yet they lost interest in what was going on around them.

Laborit urged the psychiatrists at the military hospital in Paris to try it. At 10 a.m. on 19 January 1952, a 24-year-old man in a state of severe agitation was given 50 milligrams by injection. He calmed at once, but the effect lasted only a few hours. Repeated injections irritated his veins, so on some days sleeping drugs and electroconvulsive therapy were used instead. After twenty days he was reported ready "to resume normal life". This was one patient, and other treatments were mixed in.

The people who made the drug known were Jean Delay and Pierre Deniker at the Sainte-Anne hospital in Paris. They began using it on 24 March of the same year. At first they thought of it as a cooling drug and applied ice packs; it soon became clear that cooling added nothing. For most patients 75 milligrams a day of the drug alone was enough. Within six months they had published six reports, one of them an account of 38 patients.

According to Ban, the drug did more than calm excitement. It also eased hallucinations (seeing or hearing things that are not there) and delusions (fixed beliefs in things that are not true). It went on sale in France in November 1952 and spread around the world between 1953 and 1955. A meeting in Paris in October 1955 drew 257 participants from 19 countries. The article by Jansson quoted in Part 4 says it was this drug that pushed lobotomy out of the treatment of schizophrenia.

Imipramine: the drug that failed in schizophrenia (1957)

Once chlorpromazine was selling, drug companies went looking for something similar. One was Geigy, in Switzerland, which made a series of compounds with closely related structures and asked hospital doctors to try them. One of those doctors was the psychiatrist Roland Kuhn, at the hospital in the small Swiss village of Münsterlingen.

Kuhn gave a compound labelled G22355 to patients with schizophrenia. It did little for hallucinations or delusions. It had missed its target. But he noticed other changes. Some patients became elated. And in patients whose low mood had been prominent, the low mood lifted.

Kuhn tried the compound on about a hundred patients with depression and reported the results in a Swiss medical journal on 31 August 1957. It was later named imipramine and is regarded as the first antidepressant. When he presented the work at an international congress in Zurich the following week, about a dozen people were in the audience.

Walter Brown and Maria Rosdolsky, who translated the report into English in 2015, call it extraordinarily flawed by present standards. There was no control group. There was no standard scale for rating symptoms. And Kuhn based his description of the drug's effect only on the 40 patients who had done well. Kuhn himself said his numbers were too small for statistics, and estimated that the drug had failed in a fifth to a quarter of cases. He also wrote that the effect often took one to four weeks to appear.

The next year, 1958, Kuhn summed up three years of use in more than 500 patients in an American journal. His conclusion was that the drug worked best in the type of depression that comes on without an obvious trigger.

This time, comparison groups were used

So far the three drugs had entered medicine in the same way as the treatments in Part 4: tried on a small number, reported as successful, with nothing to compare against. What differed was what came next.

Lithium was picked up by Mogens Schou and his colleagues in Denmark. Their 1954 paper opens by calling it rather astonishing that the Australian reports had drawn so little interest. It then gives three reasons why a psychiatric treatment is hard to assess. There is no yardstick for how severe the illness is. Patients and observers alike are swayed by what they expect. And mania and depression get worse and better on their own.

They gave lithium to 38 patients with mania. For some of them, lithium tablets and a placebo (a dummy tablet that looks the same but contains no drug) were swapped in a random order, usually every two weeks. The tablets arrived in boxes marked only with numbers, and neither the patients, the ward staff nor the psychiatrists knew which was which. Fourteen patients clearly improved, 18 possibly did, and six did not. Lees calls this the first attempt at a randomised controlled trial in psychiatry.

Doubts remained. In the 1960s Schou and his colleagues reported that people who kept taking lithium had fewer relapses of mania and depression. In 1968 Barry Blackwell and Michael Shepherd, in London, answered in The Lancet with a paper asking whether this was another therapeutic myth. Schou's group replied in 1970 with a trial. Eighty-four patients who had been on lithium for at least a year were matched in pairs. Within each pair, chance decided who stayed on lithium and who was switched to placebo, without the patients or their doctors knowing. Over five months, 21 of those switched to placebo relapsed, and none of those who stayed on lithium did. The United States approved lithium for mania that same year, 1970. Twenty-one years had passed since Cade's paper.

For chlorpromazine, the US National Institute of Mental Health set up a large trial. Its 1964 report opens by saying that the drug had been in use for eleven years and that there were still important questions about the extent of its clinical effects. The trial ran in nine hospitals. It allocated 463 newly admitted patients at random to chlorpromazine, one of two similar drugs, or a placebo, for six weeks. Neither doctors nor patients knew who was taking what.

A 2014 review that pooled 55 trials of this kind concluded that chlorpromazine improves symptoms overall and reduces relapse. The same review says the quality of the evidence is low, and that there are clear side effects: heavy sleepiness, tremor and stiffness, weight gain, and low blood pressure. It ends by calling the drug a well-established but imperfect treatment.

For imipramine, Britain's Medical Research Council reported a trial in 1965. Two hundred and fifty patients with depression, aged 40 to 69, were allocated at random to one of four treatments: electroconvulsive therapy, imipramine, phenelzine (an antidepressant of a different type), or a placebo. The tablets were made to look identical.

After four weeks, among those still on their allotted treatment, the share with few or no symptoms was 71 percent on electroconvulsive therapy, 52 percent on imipramine, 39 percent on placebo and 30 percent on phenelzine. At that point the gap between imipramine and placebo was no bigger than chance could produce. It became clear later. Over six months, the number who reached discharge on their allotted treatment alone was 22 of 61 on placebo (36 percent), 34 of 63 on imipramine (54 percent) and 35 of 65 on electroconvulsive therapy (54 percent). On phenelzine it was 13 of 61 (21 percent), no better than placebo.

These figures show two things that a report in Kuhn's form could not. One is that about a third of these patients got better within six months with no specific treatment. The other is that of two drugs both called antidepressants, only one worked, in these patients and at these doses. A report that only counts the people who improved cannot tell those two drugs apart.

Trials comparing insulin coma with chlorpromazine were also run during the 1950s. That is the period in which, as Part 4 described, insulin coma was "quietly forgotten".

The dopamine hypothesis: an explanation traced backwards from the drug

That the drugs worked had been checked. Why did they work?

The clue came from the drugs. Nerve cells in the brain signal to one another by passing chemicals across. On the receiving side there are receptors, docking sites that a particular chemical fits. In the late 1950s the Swedish pharmacologist Arvid Carlsson showed that one such chemical, dopamine, carries signals in the brain.

In 1963 Carlsson and his colleague Lindqvist examined the brains of mice given chlorpromazine and a related drug. Dopamine and its chemical relatives were being made and used up faster than normal. They proposed that the drugs were blocking the receptors. With the signal no longer getting through, the brain was making more of the chemical to compensate.

In the mid-1970s the idea was tested directly. In 1976 Ian Creese, David Burt and Solomon Snyder, in the United States, showed that the smaller the dose at which a schizophrenia drug works, the more strongly it binds to dopamine receptors. How well the drugs worked and how tightly they bound lined up closely.

From this came the proposal that in schizophrenia, signalling by dopamine is too strong. This is the dopamine hypothesis. But the reasoning runs backwards. Nobody studied the illness, found its cause and then designed a drug to match. A drug that worked came first, and what was going on in the illness was inferred from what the drug does. Feeling better on a medicine that brings down a fever does not show that the fever itself was the cause of the illness. The fever may be the result of a cause that lies somewhere else. The same goes for dopamine. If a drug weakens dopamine signalling and the symptoms ease, it does not follow that too much dopamine signalling is the root of the illness.

Carlsson himself said so in 1988. The dopamine hypothesis, he wrote, was still almost entirely based on pharmacological evidence. A disturbance of dopamine function in schizophrenia had not yet been established beyond doubt. Schizophrenia was probably a mixed group of disorders.

Lithium is understood even less. Lees, writing in 2020, says its mode of action remains uncertain.

What had been achieved by this point, and what had not

Two things had been achieved. One was drugs that worked. Mania, schizophrenia and depression each now had a medicine that is still used today. The other was a change in how treatments were checked. Some of the four treatments in Part 4 waited more than twenty years for a comparison group, by which time more than twenty thousand lobotomies had been done. All three of these drugs were tested against placebo, and their effects held up. The same trials also picked out what did not work.

What had not been achieved should be listed too. First, the drugs did not work for everyone. Kuhn put the failures at a fifth to a quarter. Second, side effects. Chlorpromazine affected movement and weight, and with lithium the useful dose sat close to the poisonous one.

And then the reasons. None of the three was made from an understanding of a cause; each was found on the way to something else. The explanation was assembled afterwards, from what the drug does. Twenty-five years on, the man who built that explanation wrote that it had still not been confirmed in the illness itself. The gap this series has followed since Part 1, between knowing that something works and knowing why, did not close when working drugs arrived.

The first five parts have been about Europe and the United States. The next turns to Japan, from the era when patients were confined in a room of the family home to the law of 1987.