Brain texture as a marker of transdiagnostic clinical profiles in patients with recent-onset psychosis and depression

doi:10.21203/rs.3.rs-2732176/v1

Download PDF

Article

Brain texture as a marker of transdiagnostic clinical profiles in patients with recent-onset psychosis and depression

https://doi.org/10.21203/rs.3.rs-2732176/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 02 Jan, 2024

Read the published version in Nature Mental Health →

Version 1

posted

You are reading this latest preprint version

Prediction models of brain texture changes in recent-onset psychosis (ROP) and recent-onset depression (ROD) have lately been proposed. The validation of these models transdiagnostically at the individual level and the investigation of the variability in clinical profiles are still missing. Established prevention and treatment approaches focus on specific diagnoses and do not address the heterogeneity and manifold potential outcomes of patients. We aimed to investigate the utility of brain texture changes for a) identification of the psychopathological state (ROP and ROD) and b) the association of individualized brain texture maps with clinical symptom severity and outcome profiles. We developed transdiagnostic models based on structural MRI data on 116 patients with ROD, 122 patients with ROP, and 197 healthy controls (HC) from the Personalised pROgNostic tools for early psychosIs mAnagement (PRONIA) study by applying explainable artificial intelligence and clustering analysis. We investigated the contrast texture feature as the key feature for the identification of a general psychopathological state. The discrimination power of the trained prediction model was > 72% and validated in a second independent age and sex-matched sample of 137 ROP, 94 ROD, and 159 HC. Clustering analysis was implemented to map the texture brain changes produced from an explainable artificial intelligence algorithm, in a group fashion. The explained individualized brain contrast map grouped into 8 homogeneous clusters. In each group, we investigated the association between the explained brain contrast texture map and clinical symptom severity as well as outcome profiles. Different patterns in the explained brain contrast texture map showed unique associations of brain alterations with clinical symptom severity and clinical outcomes, i.e., age, positive, negative and depressive symptoms, and functionality. In some clusters, the mean explained brain contrast texture map values and/or brain contrast texture voxels significantly contribute to the classification decision significantly predicted PANSS scores, functionality and change in functionality over time. In conclusion, we created homogeneous clusters which statistically significant predict the clinical severity and outcome profile.

Health sciences/Diseases/Psychiatric disorders/Psychosis

Health sciences/Diseases/Psychiatric disorders/Depression

Health sciences/Biomarkers/Predictive markers

Trandiagnostic psychiatry

recent onset psychosis

recent onset depression

homogeneous clusters

explainable artificial intelligence

In mental health, diagnosis and treatment depend nearly entirely on a syndromal classification that defines diagnostic categories based on symptom clusters^1–3. However, comorbidity has long been recognized as the rule rather than the exception, with half of the patients fulfilling diagnostic criteria for one disease also meeting criteria for at least one other condition^4,5. A broader, transdiagnostic approach that acknowledges the complexity of mental illnesses is essential to unravel the determinants of their psychopathology^6–8 in order to advance therapeutic discovery.

The interconnection of psychotic disorders and depressive syndromes showcases the above issue. Depressive symptoms can be observed in the prodromal stage before the first manifestation of psychotic symptoms, during the first or later psychotic episodes, as well as after the fading out of an acute episode, and as a sign of relapse⁹. Depression might result from the neurobiological processes of psychotic disorders¹⁰, as a side effect of antipsychotic drugs, an outward manifestation of problematic symptoms, or a psychological response to psychosis as a significant life experience¹¹. When individuals with clinical high risk for psychosis first present for treatment, subclinical psychotic episodes and depressive symptom levels are usually at their peak; connection of the two symptom clusters has been reported to be strongest at initial clinical presentation and gradually deteriorate over time¹². For an explanation of the connection between depression and psychosis, numerous models have been presented^13,14. Despite the growing emphasis on transdiagnostic approaches to mental health, it is unclear if distinct diagnostic categories share a common neuronal base and our knowledge is based on meta-analyses of studies that were carried out on individual disorders using different methodologies¹⁵.

Structural MRI studies have identified common brain abnormalities in patients with schizophrenia and depression (e.g., cerebral white matter¹⁶, cerebellum^17,18). Volume deficits and structural brain abnormalities have been linked to both schizophrenia^19,20 and depression^21,22. These changes have been also observed in patients with recent-onset psychosis (ROP)^19,23−26 and recent-onset depression (ROD)^8,27−29, and have a wide distribution, affecting frontal, temporal, and parietal cortical regions, as well as subcortical, cerebellar, insula and callosal regions^30–32. Moreover, white matter (WM) and cerebrospinal fluid (CSF) abnormalities have been reported in both psychotic and depressive disorders^33–36. However, it is unclear how the observed brain changes translate to clinical symptoms and how different brain change profiles correspond to different symptom profiles, i.e., whether positive and negative symptoms, depressive symptoms and functionality are associated with different brain topology.

Despite having information from structural brain imaging studies on psychosis that date back more than 40 years, so far it has not been possible to find a diagnostic or predictive biomarker for use in clinical settings. Small sample sizes, ambiguous biomarker definitions, and a dearth of replications³⁷, have been suggested to be responsible for these shortcomings. Cross-sectional studies have reported different symptom clusters (positive and negative symptoms, depressive symptoms, and functionality) that are associated with distinct neuroimaging profiles across both diagnoses^8,38 based on brain volumetric changes. In the present analysis, the contrast texture feature map, expresses the grey-levels differences of non-segmented MRIs capturing the segments inter-relation and the cortical folding, combined with the explainable artificial intelligence algorithm, giving insights regarding the association of the pattern the algorithm uses for classification with the clinical profile and outcome, in homogeneous clusters of the symptom. The current study sought to investigate a) whether a transdiagnostic set of structural alterations characterized ROP and ROD and whether any such alterations may reflect markers of psychiatric illness, and b) associations between transdiagnostically individualised brain markers with clinical symptom severity and outcome profiles. We used a large training sample of ROP and ROD and an independent validation sample, both from the Personalised pROgNostic tools for early psychosIs mAnagement (PRONIA) study³⁹. The measures of interest were radiomics texture features, based on previous findings by our group, where contrast texture feature map has been identified as the key feature for psychosis^16,38. Mathematically, the contrast feature expresses the shape abnormalities in brain cortex and the surface relief of the cortical boundary. Finally, we performed explainable artificial intelligence analysis to highlight the brain regions that most contributed to differences between ROD and ROP patients. We hypothesized that a) ROP and ROD present shared contrast texture feature abnormalities and that b) different contrast feature patterns are associated with different transdiagnostic clinical profiles.

2.1. Study Design

Data were collected in the context of the PRONIA study, an EU-FP7 funded seven-center study aiming at optimizing candidate biomarkers for the prediction of mental disorders in early states of the disease. Details of the PRONIA study sites, recruitment protocol and quality control procedures have been described in detail in previous publications^27,28.

2.2. Participants

Data were collected following the standardized recruitment and assessment protocol from the Personalised pROgNostic tools for early psychosIs mAnagement (PRONIA; https://www.pronia.eu) study across 7 European sites: Munich, Basel, Milan, Cologne, Birmingham, Turku, and Udine. Prior to being included in the study, all participants, and, in case of underage participants (defined as those under the age of 18 at all sites), and their guardians gave written informed consent. The trial was registered with the German Clinical Trials Register (DRKS00005042), and local research ethics boards in each location gave their approval.

We refer the reader to Koutsouleris et al.²⁸ for detailed description of the study design and methodology. In brief, participants were recruited through outpatient and inpatient services. The following were the general inclusion criteria: (1) age range of 15 to 40 years, (2) sufficient language skills for participation, and (3) ability to provide informed consent. Current or previous head trauma with loss of consciousness (> 5 minutes), current or past known neurological or somatic disorders potentially affecting the structure or functioning of the brain, current or past alcohol dependence, polysubstance dependence within the last 6 months, and any medical indication against MRI were all general exclusion criteria. Any current or previous DSM-IV axis I or II disorder; a positive family history for affective or nonaffective psychoses in first-degree relatives; and psychotropic medication intake more than 5 times per year and in the month prior to study inclusion were all additional HC exclusion criteria.

ROP participants had to meet the same requirements as participants for a first episode of affective or nonaffective psychotic episode, including transition criteria established by Yung et al.⁴⁰ and the Structured Clinical Interview for DSM-IV-TR (SCID)⁴¹. They also had to be less than three months removed from the start of their initial antipsychotic medication treatment. The following specific ROP exclusion criteria comprised psychosis onset occurring more than 24 months ago and antipsychotic use lasting longer than 90 days (cumulatively in the past 24 months) with a daily dose rate at or above the "First-Episode Psychosis" range minimum dosage of the German Society for Psychiatry, Psychotherapy, and Nervous Diseases (DGPPN) S3 guidelines, equivalent to 5 mg olanzapine.

Patients with ROD were required to meet SCID-established serious depression criteria that had been met within the previous three months. Any DSM-IV-TR major depressive episode that preceded the current or recent episode, the current episode’s duration of more than 24 months, or the use of antipsychotic medication for more than 30 days, as described above, were specific ROD exclusion criteria.

Patients meeting Clinical High Risk (CHR) criteria according to the ultra-high risk criteria and/or the basic symptom criterion Cognitive Disturbances²⁸ were also excluded from the ROD group. Any DSM-IV axis I or II disorder, current or past; CHR status; a history of affective or nonaffective psychoses in first-degree relatives; and use of psychotropic medications more than five times per year and in the month prior to study inclusion were the HC exclusion criteria.

2.3. Training and validation samples

The total training sample of 435 age and sex matched participants included 116 with ROD, 122 with ROP (minimally treated FEP), and 197 HC, recruited during first part of the recruitment period. All models were validated in the same participants at 12 months follow-up, as well as in an independent validation sample (137 ROD, 162 ROP and 178 HC) recruited during second part of the recruitment period. The follow-up sample consisted of 100 patients with ROP and 103 patients with ROD. Matching the validation sample to the discovery sample for sex and age resulted in a final independent validation sample of 94 ROD, 137 ROP, and 159 HC, which was used in this study (details in Figure S1). The follow-up sample consisted of 81 patients with ROP and 48 patients with ROD.

2.4. Assessments

Data for the present analysis included the following data; demographic and clinical data information (age, gender and medication exposure), Beck Depression Inventory-II (BDI-II)⁴², Positive and Negative Symptom Scale (PANSS)⁴³, Global Assessment of Functioning (global functioning-role scales)⁴⁴ and Scale of Negative Symptoms. We calculated five subscores of PANSS (positive, negative, disorganized, excitement, distress) based on the factor analysis by Wallwork et al.⁴⁵. Several factor-analytic studies have suggested that a five-factor model better captures PANSS structure in schizophrenia samples⁴⁵. For the Global Assessment of Functioning (GAF), we used ratings based exclusively on level of functioning; symptoms were not considered.

All participants underwent neuroimaging, which included sMRI. In keeping with real-world scanner heterogeneity and as part of the larger PRONIA goals, the PRONIA sites were required to (1) acquire isotropic or nearly isotropic voxel sizes of at least 1 mm resolution, (2) set the Field of View (FOV) parameters accordingly to ensure full 3D coverage of the brain, including all parts of the cerebellum, and (3) define the relaxation time (TR) and echo time (TE), as well as other ionizing radiation parameters. Supplementary Table S1 lists the scanner and parameter details of the structural MR sequences used to examine the PRONIA sample participants. See previous PRONIA report²⁷ for full MRI harmonization and data acquisition parameters.

2.5. Analysis

2.5.1 Preprocessing

All images were visually inspected, automatically defaced, and anonymised using a Freesurfer-based script prior to data centralization. Subsequently, we used the open-source CAT12 toolbox (version r1155; http://dbm.neuro.uni-jena.de/cat12/), an extension of SPM12 (http://www.fil.ion.ucl.ac.uk/spm), to segment images into gray matter (GM), WM, and CSF maps and to high-dimensionally register the segmented images to the stereotactic space of the Montreal Neurological Institute coordinates (MNI-152 space). The CAT12 toolbox was used with processing steps consisting of spatial filtering, segmentation, segmentation estimation, a local adaptive segmentation step, which adjusts the images for WM homogeneities and varying GM intensities, and a high-dimensional DARTEL registration of the image to an MNI template in the IXI database (http://www.braindevelopment.org). The non-segmented MR images, i.e., wp0*images, were used for further analysis. We used histogram equalization to adjust the contrast of a greyscale image. The original image has low contrast, with most pixel values in the middle of the intensity range. The histeq function in MATLAB produces an output image with pixel values evenly distributed throughout the grayscale range. The texture feature maps were then extracted from the transformed wp0* images (details in Supplemental Section 1.1 and Figure S2).

2.5.2 Feature extraction

We extracted texture feature maps from non-segmented images using the 2D grey-level co-occurrence matrix (GLCM) calculated in each cube (N = 3,294 feature maps in total). Based on a previous study by Korda et al.³⁸, we extracted the texture features Entropy, Sum of Entropy, Difference of Entropy, Energy, Contrast, Homogeneity. We focused on the analysis in GLCM-contrast, which gives a low weight to elements with similar grey level values, but a high weight to elements with dissimilar grey levels, indicating large differences between neighboring voxels. In our previous study, contrast feature maps were used for the identification of first-episode psychosis and clinical high risk patients resulting in the highest accuracy in previous research³⁸. Contrast captures high differences in grey levels in a small neighbourhood. The highest difference is observed in the cortical boundary and the borders between segments. Smaller differences observed in brain regions revealed the shape abnormalities not related always to volumetric changes. For this reason, we selected contrast feature maps for further investigation. The contrast of the grey-level pairs reflects intracortical myelin as has been investigated for patients with schizophrenia in low-sensory and motor areas⁴⁶.

Meaningful comparison of texture feature results between different subjects is possible, if sMRI images of the brain with similar resolution and noise levels are used, a common quantization method and the same number of grey levels in all quantized images are defined^47–50. We used voxel-by-voxel sliding 3D cube of 7x7x7 dimension as presented in previous papers^16,38. Where the boundary of the cube touches non-zero brain grey-levels, the algorithm maps the value to the center of the cube. For this reason, a 7x7x7 Gaussian kernel was used to smooth the voxel’s relevance for better localization of the results. The GLCM matrix was normalized by dividing the individual values by the total sum of the values in the matrix. The normalization was performed for each GLCM extracted in each 3D cube independently (see Supplement Figure S3 for the whole preprocessing).

All the feature maps calculated from the 2D GLCM are a function of the probability of each GLCM entry and the difference of the grey levels, g₁ and g₂⁵¹. We calculated texture feature maps only for cubes including non-zero values, as presented in Figure S3 in Supplementary file.

In order to test the transdiagnostically discrimination power of ROP and ROD against healthy controls, the registered texture feature maps on the MNI space were fed into a 20x20 nested cross-validation deep learning scheme for group classification (see in Supplement Figure S4). We focus the description of results on the contrast feature map, which gives a low weight to elements with similar grey level values, but high weight to elements with dissimilar grey levels (i.e., indicating large differences between neighboring voxels⁴⁷).

2.5.3 Classification framework

As we were interested in transdiagnostic psychopathology, ROP and ROD subjects were pooled together for individualized brain texture analysis. We fed the registered contrast texture feature maps in a multilayer neural network in MATLAB to train and cross-validate a model to discriminate transdiagnostically recent onset psychosis and depression patients from HC. Repeated nested cross-validation (rNCV) was used with 10 outer CV (CV2) permutations, 20 outer CV2 folds, 10 inner CV (CV1) permutations, and 20 inner CV1 folds. See Supplementary Figure S4 for further details. Feature selection (two-sample t-test) in the inner cycle was cross-validated by selecting a number of features appropriate to the dimension of the database, namely, the top 549 ranked features that best discriminated the 2 classes⁵². The classifier implemented was a neural network-based classifier implemented in MATLAB (MathWorks Inc., Natick, Massachusetts, USA). The network used the hyperbolic tangent sigmoid transfer function and was batch-trained using the Levenberg-Marquardt training algorithm⁵³. L2-regularization was applied to access possible types of uncertainty. We selected parameters after experimentation; 5 hidden layers (tested 2 to 7), each hidden layer consists of 200 nodes (tested 2 to 500) and 1,000 epochs. The average balanced accuracy, sensitivity and specificity were calculated across all hold-out datasets.

2.5.4 Visualization and evaluation of heatmaps

For localization of the identified brain changes between pooled ROP and ROD vs. HC, we calculated the relevance of the voxels in each class using the LRP algorithm for multilayer neural networks, as described in Bach et al.⁵⁴. LRP, in its general form assumes that the classifier can be decomposed into several layers of computation. Such layers can be parts of the feature extraction from the image or parts of a classification algorithm run on the calculated features. The output of the algorithm is an individual heatmap representing common changes in brain structure in general psychopathology. For the specific deep learning schema with 5 hidden layers and size 200, the calculation of the LRP algorithm is presented in Appendix A in Supplementary file.

The final images were visualized using the MRICron toolbox (https://people.cas.sc.edu/rorden/mricron/install.html). Visualizations of the classification results on the holdout dataset are presented in Fig. 1. The regions of interest were extracted using the AAL-VOIs atlas (https://neurovault.org/images/14257/). Additionally, we investigated cortical biomarkers for psychosis in GM, WM and CSF. We used the JHU WM tractography atlas⁵⁵ to identify the WM tractography identified by the LRP.

2.5.5 Clustering of subjects and associations with clinical variables

We implemented a clustering algorithm in order to demonstrate shared brain texture patterns across subjects. Our intention was to display the heatmap of each correctly classified subject in the independent validation sample, which was tested in the winner model of the training sample, produced by the LRP algorithm in a grouped fashion and associate the clusters with symptom severity and outcome profile⁵⁶. The affinity propagation (AP) algorithm⁵⁷, which uses the concept of information passing between the data, was selected to cluster the subject’s relevance heatmap for identifying distinct patterns of the brain changes in the transdiagnostic psychopathology. The main advantage of the AP algorithm is that the number of clusters is not predefined. The input in the clustering algorithm is a matrix M×N, where N is the number of subjects and M is the relevance of each voxel. The output of the AP algorithm is a scalar for every subject that expresses in which cluster the subject belongs driven only by the values of the brain relevance.

In addition, we performed simple and multiple regression analysis to investigate and model clinical symptoms and outcome profiles, inside the clusters, from a) the mean brain relevance (heatmap) and b) contributed voxels, in the independent validation baseline and follow-up hold out dataset.

3.1. Sample characteristics

Sociodemographic and clinical characteristics of the three participant groups (ROP, ROD and HC) for the training sample are presented in Table 1. In the training sample, there were no significant differences between ROP, ROD and HC with respect to age and sex. There were significant differences between ROP and ROD with respect to specific psychopathological symptoms and psychosocial functioning (see Table 1).

Table 1

Clinical and demographic sample details for the training sample at baseline.
Group	HC	ROD	ROP	Merged ROP-ROD	ANOVA between HC, ROP and ROD	Merged ROP-ROD in the follow-up T1
Number	197	116	122	238		203
Age mean (SD)	26.8 (6.99)	27.5 (5.98)	26.5 (5.37)	25.63 (6.14)	F(2,432) = 0.772 (P = .463 ^a)	26.94 (5.84)
Sex n (%) male	89 (29.4%)	55 (31.07%)	69 (45.7%)	124 (37.8%)	F(2,432) = 4.59 (P = .011 ^a)	111 (54.68%)
PANSS_positive (SD)	—	8.23 (2.18)	11.9 (6.26)	10.0 (4.99)	F(1,701.7) = 32.58, P = .001^*	10.03 (4.99)
PANSS_negative (SD)	—	11.6 (5.61)	14.1 (7.91)	12.8 (6.93)	F(1,322.1) = 6.91, P = .009^*	12.78 (6.92)
PANSS_excitement (SD)	—	11.4 (2.63)	14.8 (6.84)	13.0 (5.41)	F(1,616) = 23.36, P = .001^*	10.30 (3.28)
PANSS_distress (SD)	—	9.68 (2.60)	11.0 (3.77)	10.3 (3.28)	F(1,84.4) = 8.13, P = .005^*	12.94 (5.32)
PANSS_disorganization (SD)	—	12.4 (4.59)	13.5 (5.95)	12.9 (5.32)	F(1, 65.3) = 2.33, P = .13	13.04 (5.41)
SANS(SD)	—	0.506 (0.81)	0.888 (0.94)	0.692 (0.89)	F(1,195) = 9.63, P = .002^*	0.69 (0.89)
GAF (SD)	—	57.35 (14.8)	46.16 (11.93)	51.93 (14.58)	F(1,218) = 39.2, P < .001^*	65.98 (15.28)
BDI-II (SD)	—	11.008 (11.354)	9.718 (10.957)	10.3805 (11.074)	F(1,171) = 0.287, P = .593	15.96 (13.65)

Note: HC, healthy controls; ROD, recent onset depression; ROP, recent onset psychosis PANSS, Positive and Negative Symptom Scale; BDI-II, Beck Depression Inventory-II, SANS, GAF

In the validation baseline sample (age and sex matched with the training sample, details in Section 2.3), there were no significant differences between ROP, ROD and HC with respect to age and sex. There were significant differences between ROP and ROD with respect to specific psychopathological symptoms and psychosocial functioning (see Table 2).

Table 2

Clinical and demographic sample details for the independent validation sample.
Group	HC	ROD	ROP	Merged ROP-ROD	ANOVA between HC, ROP and ROD	Merged ROP-ROD in the follow-up T1
Number	159	94	137	231		132
Age mean (SD)	25.0 (5.33)	26.04 (5.47)	25.8 (6.12)	26.07 (5.71)	F(2,231) = 2.13 P =0.121	26.24 (5.74)
Sex n (%) male	70 (44.03%)	46 (48.94%)	75 (54.74%)	123 (53.25%)	F(2,233) = 0.965 P =0.382	123 (53.25%)
PANSS_positive (SD)	—	10.28 (2.95)	15.19 (4.03)	12.11 (4.14)	F(1,116) = 8107, P <.001^*	11.29 (5.48)
PANSS_negative (SD)	—	16.23 (7.19)	14.47 (6.71)	15.58 (7.05)	F(1,158) = 2.93, P = 0.089	13.38 (6.083)
PANSS_excitement (SD)	—	12.4 (3.96)	13 (4.26)	12.62 (4.08)	F(1,140) = 0.9368, P = .335^*	10.96 (3.64)
PANSS_distress (SD)	—	18.28 (6.27)	17.22 (5.48)	17.89 (5.99)	F(1,165) = 1.15, P = .221	13.78 (5.38)
PANSS_disorganization (SD)	—	13.59 (3.61)	16.97 (6.18)	14.85 (4.99)	F(1,100) = 17.85, P =<.001	13.59 (5.42)
SANS(SD)	—	1.04 (0.87)	1.35 (1.01)	1.20 (0.96)	F(1,228) = 6.30, P = .013	0.94 (1.02)
GAF (SD)	—	52.22 (12.38)	45.5 (16.36)	47.66 (15.48)	F(1,90.6) = 5.93, P <.002^*	62.62 (16.25)
BDI-II (SD)	—	24.99 (14.76)	16.413 (15.09)	20.249 (15.521)	F(1,230) = 14.36, P <.001^*	14.49 (13.02)

Note: HC, healthy controls; ROD, recent onset depression; ROP, recent onset psychosis PANSS, Positive and Negative Symptom Scale; BDI-II, Beck Depression Inventory-II, SANS, GAF

3.2. Classification results and localization

A repeated nested pooled cross-validation classifier of brain contrast texture maps achieved a balanced accuracy (BAC) of 72.43%, sensitivity 74.40% and specificity 74.45% for discrimination of patients (pooled ROD and ROP) from HC, see Table 3. 231 subjects age and sex matched were included in the independent validation sample, 139 subjects classified correctly as ROD + ROP (BAC 60.27%, sensitivity 61.2% and specificity 58.65%). In the independent validation follow-up sample 137 subjects were included, 77 subjects classified correctly and 60 of them provided all the clinical variables.

Table 3

Classification results
	Balanced Accuracy (%)	Sensitivity (%)	Specificity (%)	Follow-up ROP + ROD Prediction (%)
Training sample baseline	72.43	74.40	74.45	83.25
Independent validation sample baseline	60.27	61.2	58.65	56

Implementation of the LRP algorithm showed that voxels with the highest contribution to the difference between patients and HC (i.e., those with highest relevance for the classification decision) were located in the cerebellum, anterior and posterior midline areas, frontotemporal areas and insula right (further details in Fig. 1 and Supplementary Table S2 and S3). Additionally, we investigated cortical biomarkers for the general psychopathology in GM, WM and CSF. The regions of interest were extracted using the AAL-VOIs atlas (https://neurovault.org/images/14257/). We used the JHU WM tractography atlas⁵⁵ to identify the WM tractography identified by the LRP.

3.3. Brain relevance association with clinical symptoms and outcome profile in the validation sample

The association of the brain relevance with clinical symptoms and outcome profiles was investigated in the validation sample. Clinical variables at baseline were available for 131 patients from the validation sample (out of 139 classified correctly), and 71 patients of the follow-up sample. We tried to identify patterns in the brain relevance heatmaps which were associated with clinical symptoms severity and outcome profiles. We applied the Affinity Propagation algorithm to cluster the brain relevance heatmaps of the validation sample blindly, resulting in 8 clusters. Inside each cluster, we calculated a) SANS, GAF, and BDI total scores, as well as PANSS factor scores (positive, negative, distress, excitement, and disorganization) at baseline, see Fig. 2, and b) change scores of these variables at the 1-year follow-up, see Fig. 3. The resulting clusters are presented in Figs. 2 and 3. We summarize findings here, based on symptom sum scores as described above, and outcome profile (details in Supplementary Table S4). Cluster 1 (N = 9 at baseline and N = 6 at follow-up) presented high BDI and PANSS_negative values, improvement of PANSS_negative over time, deterioration of PANSS_disorganization and SANS, over time. In contrast to Cluster 1, Cluster 3 (N = 21 at baseline and N = 12 at follow-up) presented low BDI values and improvement of PANSS disorganization. Cluster 2 (N = 8 at baseline and N = 5 at follow-up) presented high values in GAF and low values in PANSS_negative and SANS, with high improvement of BDI and deterioration of GAF over time. In contrast to Cluster 2, Cluster 4 (N = 12 at baseline and N = 6 at follow-up) presented low values in total GAF and high values in SANS, with high improvement of GAF and SANS over time. Cluster 6 (N = 31 at baseline and N = 17 at follow-up) and Cluster 7 (N = 6 at baseline and N = 5 at follow-up) captured older and younger age, respectively. Finally, Cluster 5 (N = 22 at baseline and N = 9 at follow-up) presented deterioration in PANSS scores over time and Cluster 8 (N = 22 at baseline and N = 11 at follow-up) showed strong improvement of PANSS positive scores over time.

In clusters with a sufficient number of subjects (3, 5, 6, and 8), we repeated regression analyses to predict clinical outcomes at baseline and changes in scores between baseline and follow-up, using as predictor the mean relevance of the brain heatmap (in Table 2, the mean and standard deviation of scores of ROP and ROD in the follow-up are presented). In Cluster 3, the mean relevance map predicted improvement of PANSS score factors (positive, disorganization, distress and excitement) with negative coefficients (see Supplementary Table S5.1). In Cluster 8, the mean relevance map predicted improvement of PANSS_positive score factor with a positive coefficient (see Supplementary Table S5.2). Additionally, we tested the prediction values of the relevance of the voxels which contributed most to the classification decision. There were differential associations between voxel relevance and symptoms in the different clusters, especially for Cluster 3, 5, 6 and 8. In Cluster 3, voxels in specific brain areas which are associated with low depressive symptoms (i.e., fusiform, temporal lobe and insula from the left hemisphere, see Supplementary Table S4) predicted PANSS scores (excitement, disorganization and positive) and improvement of functionality. In Cluster 5, the brain relevance predicts the improvement of functionality from the left temporal lobe and in right hemisphere calcarine, lingual and precuneus. In Cluster 6, the brain relevance associated with high age predicts the functionality at baseline from the left occipital, temporal and fusiform gyrus. In Cluster 8, the brain relevance predicts the improvement of functionality from the left temporal lobe.

Further we calculated the volumes of the regions with predicted value to check if the contribution of the voxels to the classification decision was driven by the volumetric changes. In all clusters with high numbers of subjects (Cluster 3, 5, 6 and 8), the brain volume in the detected regions was uncorrelated with the classification relevance of the respective regions, with the exception of the precentral and frontal lobe that are not contributing to the statistical models (see Fig. 4).

In this study, we identified 8 clusters of brain heatmap patterns across patients with recent onset depression or recent onset psychosis, that showed differential associations to symptom clusters, functionality and age; i.e., clinical heterogeneity in transdiagnostic psychopathology was reflected in homogeneous clusters of contrast texture brain changes. The brain regions contributing most to the decision of our classifier included key areas implicated in psychotic and depressive disorders in studies that have assessed grey and white matter changes in patients. We used the contrast feature map and the LRP method suggested by Bach et al.⁵⁴ to train and explain a classifier for identification of the transdiagnostic psychopathology. The model showed good accuracy in classifying a combined group of ROD and ROP patients against healthy controls in the training sample, which decreased in the external validation sample. However, the innovative finding of this study is the creation of the homogeneous clusters which are associated with clinical severity and outcome profiles.

Brain alterations in hippocampus, insula, and lateral prefrontal cortex have been suggested to be critical both for emotional processing and for emergence of psychotic symptoms²¹. Changes in these areas have been observed in ROD (e.g., volume decrease in medial temporal regions)²⁸ and ROP (e.g., volume deficit in the anterior hippocampus)⁵⁸, which emphasizes the implication of brain alterations in early stages of both illnesses. GM analysis in transdiagnostic psychopathology using our method showed brain changes in regions which are reported in other studies, e.g., cerebellum²², frontal⁵⁹, temporal regions⁶⁰, cingulum⁶¹, precuneus⁶², rectus⁶³, insula⁶⁴, heschl⁶⁵, lingual⁷ and putamen left⁶⁶. The explained classification results using texture feature maps indicate posterior cingulate cortex, gyrus rectus and third ventricle (consistent with our previous results in patients with schizophrenia and major depressive disorder¹⁶), insular (consistent with previous study for clinical high risk who transitioned to psychosis and first-episode psychosis^38,67) as key regions for the transdiagnostic classification of psychopathology using the contrast feature map (further details in Table S3).

In patients with mental problems, there is mounting evidence of decreased WM function resulting in problems of synchronization and connectivity⁶⁸, with fractional anisotropy (FA) evaluated using diffusion tensor imaging being the most extensively used measure (DTI). Throughout the phases and progression of psychosis, neural alterations, particularly changes in WM connectivity, could be seen⁶⁹. There are consistent findings of significantly lower FA in the left genu of the corpus callosum, the left anterior corona radiata (ACR) and the right superior longitudinal fasciculus in ROP patients relative to healthy controls^34,70,71. Functional dysconnectivity in dorsal anterior cingulate cortex revealed in major depression⁷². Third ventricle expansion has been reported in recent onset of psychosis⁷³. Other studies have reported WM integrity abnormalities as measured by FA in major depressive disorders³⁵. The above findings are in line with our results regarding the WM changes in frontal gyrus, gyrus rectus, dorsal anterior, corpus callosum. 3rd and 4th ventricle have been reported as regions that contribute to the identification of the transdiagnostic psychopathology using the contrast texture map.

The brain relevance calculated by the LRP algorithm resulted in clusters that showed differential associations with age, positive, negative and depressive symptom clusters, and functionality. Previous studies performed sMRI-based prediction models determined social functioning in the patients with ROD²⁸. In some clusters, mean relevance values and/or relevance of voxels significantly contributing to the classification decision significantly predicted PANSS sum scores, functionality and change in functionality over time. Significant contributing voxels were located in regions that have been previously associated with these scores. In previous studies on schizophrenia and ROD, the FA of the temporal part and the temporolombic GM was positively correlated with GAF score, in consistance with findings in Cluster 6, which includes 17 ROP and 14 ROD patients at baseline^28,74. Reduction of GM in fusiform and temporal lobe were associated with PANSS scores in schizophrenia⁷⁵, as we observed in Cluster 3 which includes mainly ROP patients. Improvement of functionality has also been associated with increased GM volume in temporal lobes at baseline in a previous study⁷⁶. Our proposed method yielded further, new findings, which should be further validated in different datasets, e.g., right hemisphere calcarine, lingual and precuneus for the prediction of the functionality at follow-up, fusiform gyrus and insula for the prediction of PANSS scores at baseline and occipital and fusiform gyrus for the prediction of the GAF scores at baseline.

The innovation of this study is the development and validation of individualized transdiagnostic models which give insights regarding the relation of brain changes to clinical symptoms and functional outcome. Given that heterogeneity of the mental disorders has hindered the search for biomarkers so far^5,27,28, we created homogeneous clusters based on brain relevance patterns. We found groups with different brain alteration profiles, which a) corresponded to different transdiagnostic clinical profiles while also b) showing distinct association profiles between clinical symptom clusters and anatomy. Larger studies are warranted to investigate the optimal number of stable clusters; for the purposes of this study, it is important to show that MRI can be potentially used to form clinical categorization into more homogeneous groups than those offered by categorical classification systems^3,77,78. Distinct psychopathological profiles might be established to help distinguish patients with different syndromes, independent of other diagnostic considerations, and relate these to symptoms and clinical outcomes. Models that emphasize the role of changeable transdiagnostic illness mechanisms can help further these efforts, promoting generalizability of evidence-based treatments to routine care settings by accommodating comorbidities.

Texture feature map extraction from non-segmented brain images provides insight into voxel interrelationships of different modalities. To our knowledge, non-segmented images have never been used to detect the examined disorders due to the lack of interaction between GM, WM and CSF as an indicator of diagnosis, suggesting a novel biomarker. The main advantages of the proposed method are the interpretability of the results and the use of non-segmented images, which eliminate segmentation errors. It considers that contrast reveals microstructure changes when calculated in small 3D cubes and image intensities interrelations between GM, WM and CSF. The radiomic texture feature maps were extracted in MNI space from a registered masked T1-weighted image. The radius of the cubes and the topology of the neural network, for example, should be investigated further. For replication of results, we used the same cube size and pre-processing methods as in previous papers^16,38. Future studies should investigate a range of textural qualities, as each one reflects a different component of brain diversity. Finally, diverse methodological techniques should be examined further to acquire a better understanding of neurobiological variations across disorders and to make results useful for targeted interventions and treatment alternatives. Future studies should investigate whether several neuroimaging modalities may be combined and used for more accurate prediction.

Limitations

Results must be interpreted with a clear understanding of limitations, such as the cross-sectional and longitudinal nature of our data, which is important in interpretation because there may be dynamic and changing symptom cluster profiles that are not captured. The texture feature extraction is impacted by the variation in MRI intensity standardization. Another methodological drawback of the proposed method lies in the present lack of consensus regarding the applied image normalization method within the texture features extraction process.

The present analysis confirms that, at a symptom cluster level, the psychopathological profile of recent onset psychosis and depression follows similar patterns, that can be clustered in homogeneous groups associated with specific brain alterations. We identified grouped transdiagnostic neuroanatomical signatures, which were strongly related with homogeneous clinical severity and outcome profiles. The present findings support a hypothesis of common psychopathological profiles of recent onset psychosis and depression and recognition of the outcome profile that could inform the development of novel targeted and repurposed therapies that target specific symptom clusters rather than specific diagnoses.

Data availability

The datasets used and/or analyzed during the current study will be available from the authors upon reasonable request submitted to the PRONIA Consortium.

Acknowledgments

Computational support and infrastructure provided by the OMICS at the University of Lübeck (Germany). Preliminary results presented at ECNP conference 2022 achieved the Excellence Award.

Menchetti, M., Murri, M. B., Bertakis, K., Bortolotti, B. & Berardi, D. Recognition and treatment of depression in primary care: Effect of patients' presentation and frequency of consultation. Journal of Psychosomatic Research 66, 335-341 (2009). https://doi.org:https://doi.org/10.1016/j.jpsychores.2008.10.008
van Os, J., Pries, L.-K., ten Have, M., de Graaf, R., van Dorsselaer, S., Bak, M. et al. Schizophrenia and the Environment: Within-Person Analyses May be Required to Yield Evidence of Unconfounded and Causal Association—The Example of Cannabis and Psychosis. Schizophrenia Bulletin 47, 594-603 (2021). https://doi.org:10.1093/schbul/sbab019
Fusar-Poli, P., Solmi, M., Brondino, N., Davies, C., Chae, C., Politi, P. et al. Transdiagnostic psychiatry: a systematic review. World Psychiatry 18, 192-207 (2019). https://doi.org:https://doi.org/10.1002/wps.20631
Newman, D. L., Moffitt, T. E., Caspi, A. & Silva, P. A. Comorbid mental disorders: Implications for treatment and sample selection. Journal of Abnormal Psychology 107, 305-311 (1998). https://doi.org:10.1037/0021-843X.107.2.305
Schultze-Lutter, F., Schmidt, S. J. & Theodoridou, A. Psychopathology—a Precision Tool in Need of Re-sharpening. Frontiers in Psychiatry 9 (2018). https://doi.org:10.3389/fpsyt.2018.00446
Koutsouleris, N., Worthington, M., Dwyer, D. B., Kambeitz-Ilankovic, L., Sanfelici, R., Fusar-Poli, P. et al. Toward Generalizable and Transdiagnostic Tools for Psychosis Prediction: An Independent Validation and Improvement of the NAPLS-2 Risk Calculator in the Multisite PRONIA Cohort. Biological Psychiatry (2021). https://doi.org:https://doi.org/10.1016/j.biopsych.2021.06.023
Zhang, T., Koutsouleris, N., Meisenzahl, E. & Davatzikos, C. Heterogeneity of Structural Brain Changes in Subtypes of Schizophrenia Revealed Using Magnetic Resonance Imaging Pattern Analysis. Schizophrenia Bulletin 41, 74-84 (2015). https://doi.org:10.1093/schbul/sbu136
Lalousis, P. A., Schmaal, L., Wood, S. J., Reniers, R. L. E. P., Barnes, N. M., Chisholm, K. et al. Neurobiologically Based Stratification of Recent Onset Depression and Psychosis: Identification of Two Distinct Transdiagnostic Phenotypes. Biological Psychiatry (2022). https://doi.org:https://doi.org/10.1016/j.biopsych.2022.03.021
Sönmez, N., Romm, K. L., Andreasssen, O. A., Melle, I. & Røssberg, J. I. Depressive symptoms in first episode psychosis: a one-year follow-up study. BMC Psychiatry 13, 106 (2013). https://doi.org:10.1186/1471-244X-13-106
Häfner, H., Maurer, K., Trendler, G., an der Heiden, W., Schmidt, M. & Könnecke, R. Schizophrenia and depression: Challenging the paradigm of two separate diseases—A controlled study of schizophrenia, depression and healthy controls. Schizophrenia Research 77, 11-24 (2005). https://doi.org:https://doi.org/10.1016/j.schres.2005.01.004
Iqbal, Z., Birchwood, M., Chadwick, P. & Trower, P. Cognitive approach to depression and suicidal thinking in psychosis: 2. Testing the validity of a social ranking model. British Journal of Psychiatry 177, 522-528 (2000). https://doi.org:10.1192/bjp.177.6.522
Wigman, J. T. W., Lin, A., Vollebergh, W. A. M., van Os, J., Raaijmakers, Q. A. W., Nelson, B. et al. Subclinical psychosis and depression: Co-occurring phenomena that do not predict each other over time. Schizophrenia Research 130, 277-281 (2011). https://doi.org:https://doi.org/10.1016/j.schres.2011.03.003
Upthegrove, R. Depression in schizophrenia and early psychosis: implications for assessment and treatment. Advances in Psychiatric Treatment 15, 372-379 (2009). https://doi.org:10.1192/apt.bp.108.005629
Upthegrove, R., Marwaha, S. & Birchwood, M. Depression and Schizophrenia: Cause, Consequence, or Trans-diagnostic Issue? Schizophrenia Bulletin 43, 240-244 (2017). https://doi.org:10.1093/schbul/sbw097
Goodkind, M., Eickhoff, S. B., Oathes, D. J., Jiang, Y., Chang, A., Jones-Hagata, L. B. et al. Identification of a Common Neurobiological Substrate for Mental Illness. JAMA Psychiatry 72, 305-315 (2015). https://doi.org:10.1001/jamapsychiatry.2014.2206
Korda, A. I., Ruef, A., Neufang, S., Davatzikos, C., Borgwardt, S., Meisenzahl, E. M. et al. Identification of voxel-based texture abnormalities as new biomarkers for schizophrenia and major depressive patients using layer-wise relevance propagation on deep learning decisions. Psychiatry Research: Neuroimaging 313, 111303 (2021). https://doi.org:https://doi.org/10.1016/j.pscychresns.2021.111303
Kirschner, M., Hager, O. M., Muff, L., Bischof, M., Hartmann-Riemer, M. N., Kluge, A. et al. Ventral Striatal Dysfunction and Symptom Expression in Individuals With Schizotypal Personality Traits and Early Psychosis. Schizophrenia Bulletin 44, 147-157 (2018). https://doi.org:10.1093/schbul/sbw142
Depping, M. S., Schmitgen, M. M., Kubera, K. M. & Wolf, R. C. Cerebellar Contributions to Major Depression. Frontiers in Psychiatry 9 (2018).
Borgwardt, S. J., McGuire, P. K., Aston, J., Berger, G., Dazzan, P., Gschwandtner, U. et al. Structural brain abnormalities in individuals with an at-risk mental state who later develop psychosis. British Journal of Psychiatry 191, s69-s75 (2007). https://doi.org:10.1192/bjp.191.51.s69
Chopra, S., Fornito, A., Francey, S. M., O’Donoghue, B., Cropley, V., Nelson, B. et al. Differentiating the effect of antipsychotic medication and illness on brain volume reductions in first-episode psychosis: A Longitudinal, Randomised, Triple-blind, Placebo-controlled MRI Study. Neuropsychopharmacology 46, 1494-1501 (2021). https://doi.org:10.1038/s41386-021-00980-0
Busatto, G. F. Structural and functional neuroimaging studies in major depressive disorder with psychotic features: a critical review. Schizophrenia bulletin 39, 776-786 (2013). https://doi.org:10.1093/schbul/sbt054
Meurs, M., Groenewold, N. A., Roest, A. M., van der Wee, N. J. A., Veltman, D. J., van Tol, M.-J. et al. The associations of depression and hypertension with brain volumes: Independent or interactive? NeuroImage: Clinical 8, 79-86 (2015). https://doi.org:https://doi.org/10.1016/j.nicl.2015.03.020
Pantelis, C., Velakoulis, D., McGorry, P. D., Wood, S. J., Suckling, J., Phillips, L. J. et al. Neuroanatomical abnormalities before and after onset of psychosis: a cross-sectional and longitudinal MRI comparison. The Lancet 361, 281-288 (2003). https://doi.org:https://doi.org/10.1016/S0140-6736(03)12323-9
Crespo-Facorro, B., Roiz-Santiáñez, R., Pelayo-Terán, J. M., González-Blanch, C., Pérez-Iglesias, R., Gutiérrez, A. et al. Caudate nucleus volume and its clinical and cognitive correlations in first episode schizophrenia. Schizophrenia Research 91, 87-96 (2007). https://doi.org:https://doi.org/10.1016/j.schres.2006.12.015
Fusar-Poli, P., Radua, J., McGuire, P. & Borgwardt, S. Neuroanatomical Maps of Psychosis Onset: Voxel-wise Meta-Analysis of Antipsychotic-Naive VBM Studies. Schizophrenia Bulletin 38, 1297-1307 (2012). https://doi.org:10.1093/schbul/sbr134
Koutsouleris, N., Meisenzahl, E. M., Davatzikos, C., Bottlender, R., Frodl, T., Scheuerecker, J. et al. Use of Neuroanatomical Pattern Classification to Identify Subjects in At-Risk Mental States of Psychosis and Predict Disease Transition. Archives of General Psychiatry 66, 700-712 (2009). https://doi.org:10.1001/archgenpsychiatry.2009.62
Lalousis, P. A., Wood, S. J., Schmaal, L., Chisholm, K., Griffiths, S. L., Reniers, R. L. E. P. et al. Heterogeneity and Classification of Recent Onset Psychosis and Depression: A Multimodal Machine Learning Approach. Schizophrenia Bulletin 47, 1130-1140 (2021). https://doi.org:10.1093/schbul/sbaa185
Koutsouleris, N., Kambeitz-Ilankovic, L., Ruhrmann, S., Rosen, M., Ruef, A., Dwyer, D. B. et al. Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression: A Multimodal, Multisite Machine Learning Analysis. JAMA Psychiatry 75, 1156-1172 (2018). https://doi.org:10.1001/jamapsychiatry.2018.2165
Koutsouleris, N., Dwyer, D. B., Degenhardt, F., Maj, C., Urquijo-Castro, M. F., Sanfelici, R. et al. Multimodal Machine Learning Workflows for Prediction of Psychosis in Patients With Clinical High-Risk Syndromes and Recent-Onset Depression. JAMA Psychiatry 78, 195-209 (2021). https://doi.org:10.1001/jamapsychiatry.2020.3604
Arnone, D., McIntosh, A. M., Tan, G. M. Y. & Ebmeier, K. P. Meta-analysis of magnetic resonance imaging studies of the corpus callosum in schizophrenia. Schizophrenia Research 101, 124-132 (2008). https://doi.org:https://doi.org/10.1016/j.schres.2008.01.005
Vita, A., De Peri, L., Deste, G. & Sacchetti, E. Progressive loss of cortical gray matter in schizophrenia: a meta-analysis and meta-regression of longitudinal MRI studies. Translational Psychiatry 2, e190-e190 (2012). https://doi.org:10.1038/tp.2012.116
Lai, C.-H. & Wu, Y.-T. Frontal-insula gray matter deficits in first-episode medication-naïve patients with major depressive disorder. Journal of Affective Disorders 160, 74-79 (2014). https://doi.org:https://doi.org/10.1016/j.jad.2013.12.036
Sørensen, N. V., Orlovska-Waast, S., Jeppesen, R., Christensen, R. H. & Benros, M. E. Neuroimmunological investigations of cerebrospinal fluid in patients with recent onset depression – a study protocol. BMC Psychiatry 22, 35 (2022). https://doi.org:10.1186/s12888-021-03633-0
Walterfang, M., McGuire, P. K., Yung, A. R., Phillips, L. J., Velakoulis, D., Wood, S. J. et al. White matter volume changes in people who develop psychosis. British Journal of Psychiatry 193, 210-215 (2008). https://doi.org:10.1192/bjp.bp.107.043463
Shen, X., Reus, L. M., Cox, S. R., Adams, M. J., Liewald, D. C., Bastin, M. E. et al. Subcortical volume and white matter integrity abnormalities in major depressive disorder: findings from UK Biobank imaging data. Scientific Reports 7, 5547 (2017). https://doi.org:10.1038/s41598-017-05507-6
Giné-Servén, E., Martinez-Ramirez, M., Boix-Quintana, E., Davi-Loscos, E., Guanyabens, N., Casado, V. et al. Routine cerebrospinal fluid parameters as biomarkers in first-episode psychosis: A prospective observational study. Progress in Neuro-Psychopharmacology and Biological Psychiatry 112, 110424 (2022). https://doi.org:https://doi.org/10.1016/j.pnpbp.2021.110424
Falkai, P., Schmitt, A. & Andreasen, N. Forty years of structural brain imaging in mental disorders: is it clinically useful or not? Dialogues in Clinical Neuroscience 20, 179-186 (2018). https://doi.org:10.31887/DCNS.2018.20.3/pfalkai
Korda, A. I., Andreou, C., Rogg, H. V., Avram, M., Ruef, A., Davatzikos, C. et al. Identification of texture MRI brain abnormalities on first-episode psychosis and clinical high-risk subjects using explainable artificial intelligence. Translational Psychiatry 12, 481 (2022). https://doi.org:10.1038/s41398-022-02242-z
Koutsouleris, N., Kambeitz-Ilankovic, L., Ruhrmann, S., Rosen, M., Ruef, A., Dwyer, D. B. et al. Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression: A Multimodal, Multisite Machine Learning Analysis. JAMA Psychiatry 75, 1156-1172 (2018). https://doi.org:10.1001/jamapsychiatry.2018.2165
Yung, A. R., Phillips, L. J., McGorry, P. D., McFarlane, C. A., Francey, S., Harrigan, S. et al. Prediction of psychosis: A step towards indicated prevention of schizophrenia. British Journal of Psychiatry 172, 14-20 (1998). https://doi.org:10.1192/S0007125000297602
Regier, D. A., Kuhl, E. A. & Kupfer, D. J. The DSM-5: Classification and criteria changes. World Psychiatry 12, 92-98 (2013). https://doi.org:https://doi.org/10.1002/wps.20050
Beck AT, S. R., Brown GK. Beck Depression Inventory-II. (San Antonio, TX: Psychological Association, 1996).
Kay, S. R., Fiszbein, A. & Opler, L. A. The Positive and Negative Syndrome Scale (PANSS) for Schizophrenia. Schizophrenia Bulletin 13, 261-276 (1987). https://doi.org:10.1093/schbul/13.2.261
Andreasen, N. C. The Scale for the Assessment of Negative Symptoms (SANS): Conceptual and Theoretical Foundations. British Journal of Psychiatry 155, 49-52 (1989). https://doi.org:10.1192/S0007125000291496
Wallwork, R. S., Fortgang, R., Hashimoto, R., Weinberger, D. R. & Dickinson, D. Searching for a consensus five-factor model of the Positive and Negative Syndrome Scale for schizophrenia. Schizophrenia Research 137, 246-250 (2012). https://doi.org:https://doi.org/10.1016/j.schres.2012.01.031
Jørgensen, K. N., Nerland, S., Norbom, L. B., Doan, N. T., Nesvåg, R., Mørch-Johnsen, L. et al. Increased MRI-based cortical grey/white-matter contrast in sensory and motor regions in schizophrenia and bipolar disorder. Psychological Medicine 46, 1971-1985 (2016). https://doi.org:10.1017/S0033291716000593
Brynolfsson, P., Nilsson, D., Torheim, T., Asklund, T., Karlsson, C. T., Trygg, J. et al. Haralick texture features from apparent diffusion coefficient (ADC) MRI images depend on imaging and pre-processing parameters. Scientific Reports 7, 4041 (2017). https://doi.org:10.1038/s41598-017-04151-4
Traverso, A., Wee, L., Dekker, A. & Gillies, R. Repeatability and Reproducibility of Radiomic Features: A Systematic Review. International Journal of Radiation Oncology*Biology*Physics 102, 1143-1158 (2018). https://doi.org:https://doi.org/10.1016/j.ijrobp.2018.05.053
Buch, K., Kuno, H., Qureshi, M. M., Li, B. & Sakai, O. Quantitative variations in texture analysis features dependent on MRI scanning parameters: A phantom model. Journal of Applied Clinical Medical Physics 19, 253-264 (2018). https://doi.org:https://doi.org/10.1002/acm2.12482
Rizzo, S., Botta, F., Raimondi, S., Origgi, D., Fanciullo, C., Morganti, A. G. et al. Radiomics: the facts and the challenges of image analysis. European Radiology Experimental 2, 36 (2018). https://doi.org:10.1186/s41747-018-0068-z
Eichkitz, C. G., Amtmann, J. & Schreilechner, M. G. Calculation of grey level co-occurrence matrix-based seismic attributes in three dimensions. Computers & Geosciences 60, 176-183 (2013). https://doi.org:https://doi.org/10.1016/j.cageo.2013.07.006
Zanetti, M. V., Schaufelberger, M. S., Doshi, J., Ou, Y., Ferreira, L. K., Menezes, P. R. et al. Neuroanatomical pattern classification in a population-based sample of first-episode schizophrenia. Progress in Neuro-Psychopharmacology and Biological Psychiatry 43, 116-125 (2013). https://doi.org:https://doi.org/10.1016/j.pnpbp.2012.12.005
Lourakis, M. A Brief Description of the Levenberg-Marquardt Algorithm Implemened by levmar. A Brief Description of the Levenberg-Marquardt Algorithm Implemented by Levmar 4 (2005).
Bach, S., Binder, A., Montavon, G., Klauschen, F., Müller, K.-R. & Samek, W. On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation. PLOS ONE 10, e0130140 (2015). https://doi.org:10.1371/journal.pone.0130140
Hua, K., Zhang, J., Wakana, S., Jiang, H., Li, X., Reich, D. S. et al. Tract probability maps in stereotaxic spaces: analyses of white matter anatomy and tract-specific quantification. NeuroImage 39, 336-347 (2008). https://doi.org:10.1016/j.neuroimage.2007.07.053
Stelzer, J., Lacosse, E., Bause, J., Scheffler, K. & Lohmann, G. Brainglance: Visualizing Group Level MRI Data at One Glance. Frontiers in Neuroscience 13 (2019). https://doi.org:10.3389/fnins.2019.00972
Frey Brendan, J. & Dueck, D. Clustering by Passing Messages Between Data Points. Science 315, 972-976 (2007). https://doi.org:10.1126/science.1136800
McHugo, M., Armstrong, K., Roeske, M. J., Woodward, N. D., Blackford, J. U. & Heckers, S. Hippocampal volume in early psychosis: a 2-year longitudinal study. Translational Psychiatry 10, 306 (2020). https://doi.org:10.1038/s41398-020-00985-1
Reig, S., Moreno, C., Moreno, D., Burdalo, M., Janssen, J., Parellada, M. et al. Progression of Brain Volume Changes in Adolescent-Onset Psychosis. Schizophrenia Bulletin 35, 233-243 (2009). https://doi.org:10.1093/schbul/sbm160
Veijola, J., Guo, J. Y., Moilanen, J. S., Jääskeläinen, E., Miettunen, J., Kyllönen, M. et al. Longitudinal Changes in Total Brain Volume in Schizophrenia: Relation to Symptom Severity, Cognition and Antipsychotic Medication. PLOS ONE 9, e101689 (2014). https://doi.org:10.1371/journal.pone.0101689
Whitford, T. J., Lee, S. W., Oh, J. S., de Luis-Garcia, R., Savadjiev, P., Alvarado, J. L. et al. Localized abnormalities in the cingulum bundle in patients with schizophrenia: a Diffusion Tensor tractography study. NeuroImage. Clinical 5, 93-99 (2014). https://doi.org:10.1016/j.nicl.2014.06.003
Li, G., Rossbach, K., Zhang, A., Liu, P. & Zhang, K. Resting-state functional changes in the precuneus within first-episode drug-naive patients with MDD. Neuropsychiatr Dis Treat 14, 1991-1998 (2018). https://doi.org:10.2147/NDT.S168060
Zheng, R., Zhang, Y., Yang, Z., Han, S. & Cheng, J. Reduced Brain Gray Matter Volume in Patients With First-Episode Major Depressive Disorder: A Quantitative Meta-Analysis. Frontiers in Psychiatry 12 (2021).
Hatton, S. N., Lagopoulos, J., Hermens, D. F., Naismith, S. L., Bennett, M. R. & Hickie, I. B. Correlating anterior insula gray matter volume changes in young people with clinical and neurocognitive outcomes: an MRI study. BMC Psychiatry 12, 45 (2012). https://doi.org:10.1186/1471-244X-12-45
Dickey, C. C., McCarley, R. W., Voglmaier, M. M., Frumin, M., Niznikiewicz, M. A., Hirayasu, Y. et al. Smaller left Heschl's gyrus volume in patients with schizotypal personality disorder. Am J Psychiatry 159, 1521-1527 (2002). https://doi.org:10.1176/appi.ajp.159.9.1521
Lu, Y., Liang, H., Han, D., Mo, Y., Li, Z., Cheng, Y. et al. The volumetric and shape changes of the putamen and thalamus in first episode, untreated major depressive disorder. NeuroImage: Clinical 11, 658-666 (2016). https://doi.org:https://doi.org/10.1016/j.nicl.2016.04.008
Li, X.-B., Wang, L.-B., Xiong, Y.-B., Bo, Q.-J., He, F., Li, F. et al. Altered resting-state functional connectivity of the insula in individuals with clinical high-risk and patients with first-episode schizophrenia. Psychiatry Research 282, 112608 (2019). https://doi.org:https://doi.org/10.1016/j.psychres.2019.112608
van Velzen, L. S., Kelly, S., Isaev, D., Aleman, A., Aftanas, L. I., Bauer, J. et al. White matter disturbances in major depressive disorder: a coordinated analysis across 20 international cohorts in the ENIGMA MDD working group. Molecular Psychiatry 25, 1511-1525 (2020). https://doi.org:10.1038/s41380-019-0477-2
Carletti, F., Woolley, J. B., Bhattacharyya, S., Perez-Iglesias, R., Fusar Poli, P., Valmaggia, L. et al. Alterations in White Matter Evident Before the Onset of Psychosis. Schizophrenia Bulletin 38, 1170-1179 (2012). https://doi.org:10.1093/schbul/sbs053
Surbeck, W., Hänggi, J., Scholtes, F., Viher, P. V., Schmidt, A., Stegmayer, K. et al. Anatomical integrity within the inferior fronto-occipital fasciculus and semantic processing deficits in schizophrenia spectrum disorders. Schizophrenia Research 218, 267-275 (2020). https://doi.org:https://doi.org/10.1016/j.schres.2019.12.025
Barth, C., Lonning, V., Gurholt, T. P., Andreassen, O. A., Myhre, A. M. & Agartz, I. Exploring white matter microstructure and the impact of antipsychotics in adolescent-onset psychosis. PLOS ONE 15, e0233684 (2020). https://doi.org:10.1371/journal.pone.0233684
Ho, T. C., Sacchet, M. D., Connolly, C. G., Margulies, D. S., Tymofiyeva, O., Paulus, M. P. et al. Inflexible Functional Connectivity of the Dorsal Anterior Cingulate Cortex in Adolescent Major Depressive Disorder. Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology 42, 2434-2445 (2017). https://doi.org:10.1038/npp.2017.103
Chung, Y., Jacobson, A., He, G., van Erp, T. G. M., McEwen, S., Addington, J. et al. Prodromal Symptom Severity Predicts Accelerated Gray Matter Reduction and Third Ventricle Expansion among Clinically High-Risk Youth Developing Psychotic Disorders. Complex Psychiatry 1, 13-22 (2015). https://doi.org:10.1159/000371887
Yamada, S., Takahashi, S., Malchow, B., Papazova, I., Stöcklein, S., Ertl-Wagner, B. et al. Cognitive and functional deficits are associated with white matter abnormalities in two independent cohorts of patients with schizophrenia. European Archives of Psychiatry and Clinical Neuroscience 272, 957-969 (2022). https://doi.org:10.1007/s00406-021-01363-8
Mennigen, E., Jiang, W., Calhoun, V. D., van Erp, T. G. M., Agartz, I., Ford, J. M. et al. Positive and general psychopathology associated with specific gray matter reductions in inferior temporal regions in patients with schizophrenia. Schizophrenia Research 208, 242-249 (2019). https://doi.org:https://doi.org/10.1016/j.schres.2019.02.010
Kambeitz-Ilankovic, L., Vinogradov, S., Wenzel, J., Fisher, M., Haas, S. S., Betz, L. et al. Multivariate pattern analysis of brain structure predicts functional outcome after auditory-based cognitive training interventions. npj Schizophrenia 7, 40 (2021). https://doi.org:10.1038/s41537-021-00165-0
Haslam, N., McGrath, M. J., Viechtbauer, W. & Kuppens, P. Dimensions over categories: a meta-analysis of taxometric research. Psychological Medicine 50, 1418-1432 (2020). https://doi.org:10.1017/S003329172000183X
Markon, K. E., Chmielewski, M. & Miller, C. J. "The reliability and validity of discrete and continuous measures of psychopathology: A quantitative review": Correction to Markon et al. (2011). Psychological Bulletin 137, 1093-1093 (2011). https://doi.org:10.1037/a0025727

There is NO Competing Interest.

NatureMentalHealthSupplement.docx
Supplemental Material

Download PDF

Journal Publication

published 02 Jan, 2024

Read the published version in Nature Mental Health →

Version 1

posted

You are reading this latest preprint version

Brain texture as a marker of transdiagnostic clinical profiles in patients with recent-onset psychosis and depression

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. Method

2.1. Study Design

2.2. Participants

2.3. Training and validation samples

2.4. Assessments

2.5. Analysis

2.5.1 Preprocessing

2.5.2 Feature extraction

2.5.3 Classification framework

2.5.4 Visualization and evaluation of heatmaps

2.5.5 Clustering of subjects and associations with clinical variables

3. Results

3.1. Sample characteristics

3.2. Classification results and localization

3.3. Brain relevance association with clinical symptoms and outcome profile in the validation sample

4. Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1