Fading Hippocampal Connections May Signal Worsening Symptoms Before Psychosis Strikes

A subtle weakening of communication inside the hippocampus, the brain’s seahorse-shaped memory hub, may offer clinicians one of the first measurable warnings that a young person at risk of psychosis is heading toward a darker clinical path. That is the central conclusion of a new study published in Nature Mental Health, which tracked hundreds of individuals at clinical high risk for psychosis and found that declining connectivity within the hippocampus foreshadowed worsening negative symptoms, depressive symptoms and day-to-day functioning. Crucially, the signal appeared before the symptoms themselves deteriorated, raising the tantalizing possibility that a routine brain scan could one day help sort vulnerable young people according to the trajectory they are likely to follow.

The research, led by Lukas Roell of the University of Melbourne and LMU Munich together with colleagues across Australia, Germany and Brazil, addresses one of the most stubborn gaps in modern psychiatry. Psychotic disorders such as schizophrenia still lack treatment-informative biomarkers, and this deficit is most consequential during the earliest stages of illness, when interventions are thought to be most effective. Clinicians can identify people at clinical high risk for psychosis using structured interviews that detect attenuated, subthreshold psychotic symptoms, but the field has struggled to predict which of these individuals will deteriorate, which will recover, and which will develop full-blown psychosis. A reliable neural marker could transform that uncertainty into actionable risk stratification.

To hunt for such a marker, the team turned to the North American Prodrome Longitudinal Study, known as NAPLS-3, a multicenter observational cohort that has followed young people at clinical high risk for psychosis with repeated clinical assessments and magnetic resonance imaging. The analysis drew on longitudinal clinical and functional neuroimaging data from 434 participants, comprising 356 individuals at clinical high risk and 78 healthy controls, collected across an eight-month period. Rather than taking a single snapshot, the investigators modeled how connectivity changed over time within each participant and how those changes related to the parallel evolution of symptoms and functioning, using latent variable regression models that can accommodate measurement error in both the imaging and clinical variables.

The target of the analysis was not the hippocampus’s famous long-range connections to cortical networks, but its internal wiring. Functional connectivity measured with resting-state functional MRI reflects the statistical synchronization of activity between brain regions, and the hippocampus, despite its small size, is organized along its long axis into anterior, intermediate and posterior segments with distinct connectivity profiles. By examining connectivity within the hippocampus itself, the researchers sought to integrate two influential strands of etiological theory: one pointing to hippocampal pathology as a core feature of schizophrenia, and the other, the dysconnection hypothesis, framing psychotic disorders as disorders of neural communication rather than of isolated brain regions.

The results were strikingly specific. Decreases in intrahippocampal connectivity over the eight-month window tracked worsening negative symptoms, the amotivational and socially withdrawn features of the psychosis spectrum that are notoriously difficult to treat, as well as worsening depressive symptoms and declining psychosocial functioning. The same relationship did not hold for attenuated positive symptoms, such as unusual thoughts or perceptual disturbances, nor for cognition. The pattern was also specific to the high-risk group: healthy controls showed no comparable link, and the association did not emerge when the researchers examined connectivity within other brain areas. In other words, this was not a generic signature of scanning noise or of general distress, but a localized, diagnostically relevant signal.

Perhaps the most consequential finding concerns the direction of time. By modeling the temporal sequence of the associations, the team found that an early decrease in connectivity within the hippocampus preceded a subsequent worsening of negative symptoms, rather than the other way around. This ordering matters enormously for interpretation. If symptoms had predicted the connectivity decline, the brain change could simply be a downstream consequence of illness behavior, such as social withdrawal or poor sleep. Because the neural change came first, it is a credible candidate for a predictive marker, one that could flag deteriorating trajectories before they become clinically obvious and before opportunities for early intervention have slipped away.

Yet the study also delivers a sobering caveat. Intrahippocampal connectivity decline did not predict transition to psychosis, the outcome that has historically dominated research on the clinical high-risk state. This null result reframes what the marker is actually for. Most young people identified as high risk never develop psychosis, but many still experience significant affective and functional difficulties, and recent meta-analytic work has emphasized the transdiagnostic burden carried by this population, including high rates of comorbid depression and anxiety. The new findings suggest that hippocampal connectivity is less a crystal ball for psychotic conversion than a barometer of the affective-motivational and functional trajectories that shape quality of life regardless of whether psychosis ever emerges.

The specificity of the signal also fits a growing body of evidence implicating the hippocampus in the earliest phases of psychotic illness. Prior studies have documented resting hyperperfusion of the hippocampus in people at ultra-high risk, aberrant interactions between hippocampal activity and striatal dopamine in clinical high-risk individuals, and postmortem evidence of reduced hippocampal neuron density and oligodendrocyte numbers in schizophrenia. The hippocampus sits at the nexus of risk and resilience for the disorder, and its internal circuitry, organized along a long axis with functionally differentiated subregions, offers a plausible substrate for the motivational and affective disturbances captured by negative symptom ratings. The new study extends this literature by showing that the relevant signal is longitudinal, internal to the structure, and temporally predictive.

Methodologically, the work leans on the scale and rigor of NAPLS-3, whose methods and baseline description were published in 2022, and on modern neuroimaging pipelines, including the fMRIPrep preprocessing workflow and parcellation schemes derived from the Human Brainnetome Atlas. The authors made their analysis code publicly available on GitHub, and the underlying data are accessible through the NIMH Data Archive, subject to data access permissions. These choices matter because functional connectivity measures have historically been criticized for modest test-retest reliability, and longitudinal designs that track within-person change over months place heavy demands on measurement stability. Replication across independent cohorts, and eventually within individuals using precision functional mapping, will be essential before the marker can move into clinical use.

The translational horizon is already visible. The authors suggest that early reductions in intrahippocampal connectivity may help stratify at-risk individuals according to their expected affective and functional outcomes, and they point to neurostimulation as a potential intervention target, noting emerging work on lesion-derived psychosis circuits and on holographic transcranial ultrasound neuromodulation capable of recruiting distributed brain circuits. Large-scale efforts such as the Accelerating Medicines Partnership schizophrenia cohort study are simultaneously building the infrastructure for biomarker-driven care in the high-risk state. If the findings hold, a clinician seeing an anxious, withdrawn teenager with attenuated psychotic symptoms might one day order a resting-state scan, and a decline in the hippocampus’s internal dialogue could tip the balance toward earlier, more targeted support, long before the most damaging symptoms take hold.

Subject of Research: Intrahippocampal functional connectivity as a longitudinal neuroimaging marker of early clinical trajectories in the psychosis risk state

Article Title: Connectivity within the hippocampus as a neural marker of early clinical trajectories in the psychosis risk state

Article References: Roell, L., Lindner, C., Tian, Y. E., Chopra, S., Maurus, I., Moussiopoulou, J., Yakimov, V., Korman, M., Keeser, D., Schmitt, A., Falkai, P., Di Biase, M. A., Zitzmann, S., & Zalesky, A. (2026). Connectivity within the hippocampus as a neural marker of early clinical trajectories in the psychosis risk state. Nature Mental Health. https://doi.org/10.1038/s44220-026-00739-w

Image Credits: AI Generated

DOI: 10.1038/s44220-026-00739-w

Keywords: hippocampus, psychosis risk, clinical high risk, functional connectivity, negative symptoms, NAPLS-3, biomarker, schizophrenia, neuroimaging, depression, early intervention, neurostimulation