A new kind of pill for bipolar mania beats a dummy pill by a narrow margin

In Short. In a three-week hospital trial, a pill that works on the brain in a new way eased mania slightly more than a dummy pill. More patients on the drug left early, so larger trials must confirm it.

BIPOLAR MANIA is a phase of bipolar disorder in which a person's mood, energy and impulsive behavior rise so high that treatment often has to take place in hospital. The drugs used for it, such as lithium, valproate and antipsychotics, do work, but their side effects lead many patients to stop taking them. RAP-219, a once-daily pill made by Rapport Therapeutics, reduces the activity of a receptor for glutamate, the brain's main excitatory chemical messenger, but only in regions such as the hippocampus and neocortex, where that receptor is paired with a helper protein called TARPγ8.

On October 20th Rapport reported that the pill eased mania more than a dummy pill did in a three-week trial of 253 adults treated in hospital in America (Rapport release). The trial is also a good example of how to read a study of a psychiatric drug.

The trial relied on the Young Mania Rating Scale. A clinician interviews the patient and scores 11 symptoms, among them mood, sleep, speech, irritability and aggressive behavior, as they were over the previous 48 hours. Totals run from 0 to 60, and higher scores mean more severe mania. Patients needed a score of at least 25 to enter, and the average at the start was 36.9 in the drug group and 35.8 in the comparison group. Before the trial began, the company fixed the change on this scale after three weeks as its main test of success, its primary endpoint (trial registry).

Patients were assigned to their treatment by chance, or randomized, so that the two groups would differ in nothing except the pill. Some 126 people received RAP-219 at 0.75mg a day and 127 received an inert pill made to look the same, the placebo. Neither the patients nor the doctors rating them knew who was taking which.

After three weeks, scores had fallen by 15.4 points on the drug and by 12.3 points on placebo (Rapport presentation). The placebo group therefore improved by about a third of its starting score (12.3 of 35.8 points). Patients in both groups were in hospital and under daily care, and both could be given lorazepam, a sedative, when their symptoms required it. A review of 38 earlier mania trials by Aysegul Yildiz and colleagues, published in 2011, also found that trials run at more sites showed larger placebo effects (Yildiz et al.). Without a placebo group, the whole fall of 15.4 points might have been credited to the drug. The part that can be attributed to RAP-219 is the difference between the groups, which was 3.1 points.

Whether a gap of that size reflects the drug or chance is judged with the p-value, the probability of seeing a gap at least this large if the drug in fact did nothing. Here it was 0.045, just below the conventional cutoff of 0.05. The p-value does not say how large the true benefit is, or whether 3.1 points on a 60-point scale is a change that patients would notice. Judging that requires the spread of individual scores, which Rapport has not yet released, so a comparison with the average benefit of established mania drugs in the 2011 review will have to wait for fuller data.

The figure that most deserves attention is the number of patients who left. Across the 36 trials in the 2011 review that reported it, patients on placebo dropped out more often on average than those on drugs (42.6% against 34.2%). In this trial the pattern was reversed. Some 31 of 126 patients on RAP-219 (24.6%) stopped treatment early, against 15 of 127 on placebo (11.8%). Most withdrew their consent (21 on the drug and 6 on placebo). Rapport's analysis assumed that the missing scores of those who left would have resembled those of similar patients who stayed. If the patients who left the drug group had been doing worse, that assumption would make the drug look better than it is, and the trial cannot show whether they were.

Side effects were reported by about half of each group (65 of 126 on the drug and 67 of 127 on placebo), and none were severe. Rapport plans to discuss the results with the Food and Drug Administration in the first quarter of 2027. It expects to start two larger trials in the second half of that year, with a higher dose of 1.25mg and with sites outside America. Those trials will show whether the gap of 3.1 points holds up when more patients take part.