Structure/function interrelationships and illness insight in patients with schizophrenia: a multimodal MRI data fusion study

Marie‐Luise Otte, Mike M. Schmitgen, Nadine D. Wolf, Katharina M. Kubera, Vince D. Calhoun, Stefan Fritze, Lena S. Geiger, Heike Tost, Ulrich Seidl, Andreas Meyer‐Lindenberg, Dušan Hirjak, Robert Christian WolfEuropean 2023 · DOI

TL;DR

Researchers used a multimodal MRI "data fusion" technique to discover that a lack of illness insight in schizophrenia is tied to structural and functional deficits in specific brain networks. The study reveals that the inability to recognize one's own mental health condition is not just a psychological defense, but is deeply rooted in the neural systems responsible for cognitive control, memory, and self-reflection.

What Problem Does This Paper Address?

One of the most challenging aspects of treating schizophrenia (SZ) is "lack of insight"—a condition where patients do not recognize they have a mental disorder or that their experiences (like hallucinations) are symptoms of an illness. This isn't simple denial; it’s a clinical phenomenon associated with poorer treatment outcomes, higher relapse rates, and lower quality of life.

While previous research identified specific brain regions linked to insight, most studies looked at brain structure (anatomy) or function (activity) in isolation. This paper addresses the gap by investigating the interrelationship between structure and function. The authors sought to determine if the "strength" of specific neural networks—rather than just isolated spots—could explain why insight varies so significantly among patients.

Methods

The researchers employed a sophisticated multivariate approach called Parallel Independent Component Analysis (pICA).

  • The "Fusion" Aspect: Unlike standard analyses that look at one data type at a time, pICA allows researchers to "fuse" two different MRI modalities to find linked patterns.
  • Structural Data: They used Voxel-Based Morphometry (VBM) to measure gray matter volume (GMV), providing a map of brain anatomy.
  • Functional Data: They calculated the Amplitude of Low Frequency Fluctuations (ALFF) from resting-state fMRI. This measures the "intensity" of spontaneous brain activity while a participant is at rest.
  • The Goal: pICA identifies "components" (networks of brain regions) that vary together across the participant group in both structure and function, then correlates these network strengths with clinical scores of illness insight.

Datasets

The study involved a total of 121 participants:

  • 74 Patients with Schizophrenia: Recruited from the Central Institute of Mental Health in Mannheim, Germany. All were diagnosed with the paranoid subtype and were on stable medication.
  • 47 Healthy Controls: Matched for demographics, with no personal or family history of mental disorders.
  • Clinical Assessment: Insight was measured using the Osnabrück Scale of Adherence and Identification (OSSTI), which distinguishes between "identifying symptoms" and "accepting the need for treatment."

Key Insights and Ideas

  • Insight is a Network Property: The study confirms that illness insight isn't localized to a single "insight center." Instead, it emerges from the integrity of large-scale networks, particularly the frontoparietal network (involved in goal-directed behavior) and the Default Mode Network (DMN) (involved in self-referential thought).
  • Structural vs. Functional Links: Interestingly, the strongest links to insight were found in structural networks. Reduced gray matter volume in the frontoparietal and posterior cortical regions was significantly associated with lower insight.
  • The Medial Temporal Connection: A functional network involving the superior temporal gyrus was also linked to insight. This suggests that the way the brain retrieves memories and puts thoughts into context—functions of the medial temporal lobe—is critical for a patient to "label" a psychotic experience as a symptom rather than reality.
  • Independence from General Cognition: The researchers found that insight scores did not strongly overlap with general neurocognitive tests (like memory or fluency). This supports the idea that illness insight is a distinct clinical domain, not just a byproduct of general "brain fog" or cognitive decline.

Key Results

  • Group Differences: The study identified 12 distinct networks (7 structural, 5 functional) that differed significantly between SZ patients and healthy controls, primarily in frontoparietal, temporal, and cerebellar regions.
  • Structural Correlations: Lower insight (OSSTI scores) was significantly correlated with structural deficits in:
  • Frontoparietal Network: Including the pre- and postcentral gyrus, thalamus, and precuneus ($r = -0.28$).
  • Posterior Cortical Network: Including the cuneus, lingual gyrus, and posterior cingulate ($r = -0.26$).
  • Functional Correlations: Lower insight was associated with aberrant activity in a Medial Temporal Network ($r = -0.26$).
  • Clinical Link: A positive correlation was found between Global Assessment of Functioning (GAF) and treatment adherence, reinforcing that better insight translates to better real-world functioning.

Limitations and Future Work

  • Cross-Sectional Design: The study captures a "snapshot" in time. Because insight can fluctuate during the course of the illness, the authors suggest that future studies should use longitudinal designs or "Ecological Momentary Assessments" (smartphone-based tracking) to see how these networks change as insight improves or worsens.
  • Statistical Stringency: While the correlations were significant at a nominal level ($p < 0.05$), they did not survive the highly conservative Bonferroni correction for multiple comparisons. The authors advise interpreting these specific associations with "appropriate caution."
  • Specificity: Because the study compared patients to healthy controls, some findings might reflect general schizophrenia-related brain changes rather than markers exclusive to insight. Future work comparing "high-insight" vs. "low-insight" patient groups directly could further isolate these neural signatures.

Summary updated

Abstract

Illness insight in schizophrenia (SZ) has an important impact on treatment outcome, integration into society and can vary over the course of the disorder. To deal with and treat reduced or absent illness insight, we need to better understand its functional and structural correlates. Previous studies showed regionally abnormal brain volume in brain areas related to cognitive control and self-reference. However, little is known about associations between illness insight and structural and functional network strength in patients with SZ. This study employed a cross-sectional design to examine structural and functional differences between patients with SZ (n = 74) and healthy controls (n = 47) using structural and resting-state functional magnetic resonance imaging (MRI). Voxel-based morphometry was performed on structural data, and the amplitude of low frequency fluctuations (ALFF) was calculated for functional data. To investigate abnormal structure/function interrelationships and their association with illness insight, we used parallel independent component analysis (pICA). Significant group (SZ vs. HC) differences were detected in distinct structural and functional networks, predominantly comprising frontoparietal, temporal and cerebellar regions. Significant associations were found between illness insight and two distinct structural networks comprising frontoparietal (pre- and postcentral gyrus, inferior parietal lobule, thalamus, and precuneus) and posterior cortical regions (cuneus, precuneus, lingual, posterior cingulate, and middle occipital gyrus). Finally, we found a significant relationship between illness insight and functional network comprising temporal regions (superior temporal gyrus). This study suggests that aberrant structural and functional integrity of neural systems subserving cognitive control, memory and self-reference are tightly coupled to illness insight in SZ.
BibTeX
@inproceedings{otte2023structure,
  author = {Marie‐Luise Otte and Mike M. Schmitgen and Nadine D. Wolf and Katharina M. Kubera and Vince D. Calhoun and Stefan Fritze and Lena S. Geiger and Heike Tost and Ulrich Seidl and Andreas Meyer‐Lindenberg and Dušan Hirjak and Robert Christian Wolf},
  title = {{Structure/function interrelationships and illness insight in patients with schizophrenia: a multimodal MRI data fusion study}},
  booktitle = {European},
  year = {2023},
  doi = {10.1007/s00406-023-01566-1},
}
Figure