Psychiatrists weigh screening psychosis patients for antibodies that attack the brain

The thing this reporting doesn't tell you is the denominator. Around 3 percent of people will have a psychotic episode in their lifetime ^[10]. Autoimmune encephalitis is described as a rare form of brain inflammation ^[3]. According to Scientific American, the suspicion among researchers is that psychiatry is overlooking "a small number" of patients like Jones's ^[7]. The article does not estimate how many of those psychotic episodes are autoimmune.

Most of what is known about how these cases present traces back to a 2007 paper describing 12 women with autoimmune encephalitis ^[12]. Their antibodies bound to NMDA receptors, proteins involved in learning and in forming memories ^[13]. Nine of the 12, or 75 percent, first showed psychosis-like symptoms ^[1]. Eleven of the 12 went on to have seizures ^[2]. Tumors appeared to be the trigger, 90 percent of them ovarian, and all five tumors the researchers examined contained nerve tissue carrying NMDA receptors ^[14].

So in that series, all but one patient developed a clear neurological sign. Jones's patient had no obvious neurological symptoms, and by Scientific American's account the case would likely have been missed had Jones not suspected an autoimmune cause ^[6]. The signs that usually firm up the diagnosis are bright patches of inflammation on MRI, psychosis that worsens quickly, and neurological symptoms such as seizures ^[5]. Without the seizures, more of the diagnosis rests on the antibody test itself, run on blood or spinal fluid ^[4]. Patients in the early stages often show hallucinations, delusions and agitation, and are first sent to psychiatric services ^[11].

The case itself is a sequence of events in one person. He stroked and interacted with the animals he saw "in as real a way as he was with me," Jones said ^[1]. Antipsychotics did nothing for the visions. After plasma exchange, they dissipated ^[2]. One patient with no comparison cannot separate the effect of a treatment from the natural course of an illness. The broader claim in the reporting is that psychotic symptoms often disappear when these patients get drugs that suppress the immune system ^[3].

Researchers are now considering screening a wider range of psychiatric patients and hunting for antibodies not yet tied to the condition ^[7]. The same reporting records the opposing worry: overdiagnosis could plant false hopes and divert patients from the psychiatric treatment the vast majority will actually need ^[8]. Michael Zandi, a neurologist at University College London, said the "artificial divide" between psychiatry and neurology is starting to break down ^[9].

I think this evidence supports ordering an antibody test for a patient who looks like Jones's: psychosis that does not respond to antipsychotics, or that worsens fast. It does not yet support testing everyone who presents with psychosis. That would take two figures from a screened psychiatric population without neurological signs: the share who test positive, and how many of those improve on immunotherapy. A negative result on today's panel also has limits. Most cases are linked to a brain-targeting autoantibody, and the triggers range from tumors to earlier infections to causes nobody has identified ^[15].