Schizophrenia's synapse loss follows a different map than MRI shows

For the first time in living patients, researchers have directly measured where schizophrenia strips the brain of its connection points. The loss is pronounced, organized, and skews to the left side of the brain, with a left frontal region emerging as a possible starting point. The pattern is also not where the brain physically shrinks.

The finding, published in Molecular Psychiatry, comes from 29 people with schizophrenia and 93 healthy controls scanned with [11C]UCB-J PET, a specialized brain scan that lights up a protein sitting on synapses, the tiny junctions where brain cells pass signals to one another. Until now, the field had inferred synapse loss only indirectly, through postmortem tissue or anatomical MRI that measures the size of brain structures. The new data put two maps side by side, where synapses are lost and where the brain's grey matter visibly thins, and they do not line up.

The asymmetry came through first. Across regions, the left hemisphere took a harder hit than the right, with a Cohen's D effect size of 1.14 (p < 0.001) for left-side regions versus the right. Per-region effect sizes ranged from 0.58 to 1.47 Cohen's D, all surviving family-wise error correction at pFWE < 0.05. The loss reached into frontal, temporal, cingulate, thalamic, striatal, and hippocampal areas, a wide footprint, but not a uniform one.

The synaptic-loss pattern lines up with two other maps. It tracks the brain's molecular architecture, including where dopamine and glutamate receptor systems are concentrated, and the brain's intrinsic functional networks, the webs of regions that tend to fire together at rest. It does not track the pattern of grey matter shrinkage that decades of MRI studies have charted in schizophrenia. When clinicians have been pointing MRI at the brain's anatomy, they have been reading the wrong proxy. The chemistry is the more sensitive readout. Synapse loss in schizophrenia is following the chemistry, not the anatomy.

The researchers, led by first author Sidhant Chopra (now at Orygen and the University of Melbourne) with senior authors Avram Holmes at Rutgers and Rajiv Radhakrishnan at Yale, also point to one region in the front-left of the brain as a candidate starting point for the cascade. The paper calls it a candidate, not a proven origin, leaving the finding as a hypothesis to test rather than a settled map. The Rutgers press release on July 14 and ScienceDaily's coverage on August 20 describe a study that has been in press since the journal's online publication.

The patient subgroup is small: 29 people with schizophrenia, even though the study is one of the largest in-vivo synaptic-density PET comparisons in the field to date. [11C]UCB-J PET is a research instrument available at only a handful of imaging centers worldwide, and no one is using it to diagnose or stage schizophrenia in clinics. The study is correlational. It shows where synapses are reduced, not that the loss causes the symptoms. The candidate-region language is the researchers' own.

Future trials can target the implicated regions and circuits with pharmacology or stimulation, rather than casting a wide net across the cortex, and use synaptic-density PET as a biomarker to read out whether the intervention moved the needle. Decoupling the two maps gives the field somewhere concrete to look next. It also gives the older postmortem findings of synaptic protein loss in schizophrenia a spatial target to test against in living patients.

The next test is replication in a larger patient cohort and a longitudinal design that follows people over time to see whether the left frontal candidate region is, in fact, where the cascade starts. The team at Rutgers and Yale is part of a small consortium of groups with [11C]UCB-J PET capability, and coordinating multi-site scans will be the rate-limiting step for any clinical translation.