Gyrification Across Psychotic Disorders: A Bipolar-Schizophrenia Network of Intermediate Phenotypes Study

Abstract

Background:

The profiles of cortical gyrification across schizophrenia, bipolar disorder I, and schizoaffective disorder have been studied to a limited extent, report discordant findings, and are rarely compared in the same study. Here we assess gyrification in a large dataset of psychotic disorder probands, categorized according to the DSM-IV. Furthermore, we explore gyrification changes with age across healthy controls and probands.

Methods:

Participants were recruited within the Bipolar-Schizophrenia Network of Intermediate Phenotypes study and received T1-MPRAGE and clinical assessment. Gyrification was measured using FreeSurfer 7.1.0. Pairwise T-tests were conducted in R, and age-related gyrification changes were analyzed in MATLAB. P values < 0.05 after false discovery rate correction were considered significant.

Results:

Significant hypogyria in schizophrenia, bipolar disorder, and schizoaffective disorder probands compared to controls was found, with a significant difference bilaterally in the frontal lobe between schizophrenia and bipolar disorder probands. Verbal memory was associated with gyrification in the right frontal and right cingulate cortex in schizophrenia. Age-fitted gyrification curves differed significantly among psychotic disorders and controls.

Conclusions:

Findings indicate hypogyria in DSM-IV psychotic disorders compared to controls and suggest differential patterns of gyrification across the different diagnoses. The study extends age related models of gyrification to psychotic disorder probands and supports that age-related differences in gyrification may differ across diagnoses. Fitted gyrification curves among probands categorized by DSM-IV significantly deviate from controls, with the model capturing early hypergyria and later hypogyria in schizophrenia compared to controls; this suggests unique disease and age-related changes in gyrification across psychotic disorders.

1. Introduction

Gyrification is the process through which folds in the cerebral cortex are formed; it occurs at a critical stage during fetal neural development and organization (White et al., 2009; Hogstrom et al., 2013) and it is thought to reflect underlying neural connectivity (White et al., 2009; White and Hilgetag, 2011). Gyrification can be characterized using the Local Gyrification Index (LGI), a surface-based method using 3-D cortical reconstruction to compute local ratios of the inner surface to the outer smoothed cortical surface across the brain (Schaer et al., 2008). Gyrification is thought to reflect underlying neural connectivity (White et al., 2009) as theories have proposed neuronal migration during fetal development in the second trimester produces differential fiber tensions that interact globally across the cortical surface to form gyri and sulci (White and Hilgetag, 2011). Abnormalities in connectivity have been observed across psychosis spectrum disorders, such as schizophrenia (SZ), psychotic bipolar (BP), and schizoaffective disorders (SAD). Hence, investigating cortical gyrification is critical to understand the underlying development of psychosis spectrum disorders. Yet, few studies have investigated abnormalities of gyrification in psychotic disorders.

Previous studies in SZ reveal abnormal gyrification in comparison to healthy controls (HC); however, the findings are inconsistent across studies, with some studies reporting decreased gyrification, called “hypogyria”; increased gyrification, called “hypergyria”; or no differences across the same hemispheres and cortical lobes for psychotic disorder probands compared to controls (Matsuda and Ohi, 2018). Intriguingly, one study has found a change from hypergyria to hypogyria in Broca’s region and the left insula after a 2-year follow-up (Palaniyappan et al., 2013); this raises the question if there is an altered neurodevelopmental trajectory of gyrification changes in SZ across the whole brain. Similar heterogeneous results are found in studies of gyrification in BP probands (Miola et al., 2022). Only a handful of studies (Nanda et al., 2013; Vogeley et al., 2001) have investigated gyrification in SAD, including a previous investigation covering the first wave of data from the Bipolar-Schizophrenia Network of Intermediate Phenotypes (BSNIP) consortium. Nanda et al. (2013) reported hypogyria in the frontal, parietal and temporal lobes; meanwhile, Vogeley et al. (2001), in analysis of post-mortem brains, reported right frontal hypergyria in male (N=11) but not in female (N=13) probands, attributable to different experimental set up and small sample size.

With respect to development, age-related decrements in gyrification have been identified in healthy adults (Hogstrom et al., 2013; Magnotta et al., 1999), but the impact of age on cortical gyrification in probands with psychosis spectrum disorders is unclear. Though the literature broadly reports discordant results for psychotic disorders, numerous studies on first episode psychosis or young, non-chronic probands with SZ report hypergyria across various cortical regions in their samples compared to controls (Matsuda and Ohi, 2018; Schultz et al., 2013; Del Casale et al., 2021; Sasabayashi et al., 2020; Narr et al., 2004; Harris et al., 2004; Sasabayashi et al., 2021), while studies of probands with chronic SZ often report hypogyria compared to controls (Matsuda and Ohi, 2018; Sasabayashi et al., 2021; Cao et al., 2017; Nesvåg et al., 2014; Sallet et al., 2003; Kitajima et al., 2023). Nevertheless, a central question of why there are discordant results within the gyrification literature is still unanswered. Given the large cross-sectional dataset of the BSNIP consortium that captured a broad age-range of probands from 18 to 60 years, we aimed to create a mathematical model that captured the relationship between gyrification and age within the sample; we hypothesized, based off of the previous discordant results, that the resultant curves would capture both hyper- and hypogyria for SZ probands compared to healthy controls.

We employ the Bipolar-Schizophrenia Network on Intermediate Phenotypes (B-SNIP) consortium dataset that encompasses SZ, BP, and SAD probands, a transdiagnostic multisite collaboration formed to understand the underlying biology of the psychotic disorders. The aim of this study was to (i) identify and compare changes in LGI across psychotic disorders; (ii) investigate LGI patterns specific to each categorization; (iii) investigate the relationship between LGI and cognitive symptomatology across psychotic disorders, and (iv) model age-related gyrification changes in DSM-IV categorized probands and healthy controls. We hypothesized that (1) psychotic disorder probands would have lower gyrification than controls; (2) higher gyrification for BP probands would be observed in comparison to SZ probands; (3) lower gyrification would be correlated with worse cognitive symptomatology across psychotic disorders; and (4) there will be an altered gyrification curve between SZ and healthy control probands such that both hypergyria and hypogyria will be observed for SZ compared to control in the mathematical model, based off of previous discordant results reported in the literature.

2. Methods and Materials

2.1. Sample

Participants (N = 1,860) were recruited from the B-SNIP consortium: B-SNIP 1, B-SNIP 2, and PARDIP (Psychosis and Affective Research Domains and Intermediate Phenotypes) studies. Non-psychosis controls (HC) were recruited alongside participants with SZ (N = 440), SAD (N = 350), or BP (N = 343). No participants within this study were related. Participants who used any substances within six months of the study were excluded. Participants were assessed with the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders-IV Axis I Disorders (SCID DSM-IV). Tamminga et al. (2012) describes in further detail the recruitment inclusion and exclusion criteria.

2.2. Imaging

Participants received a 3T-MPRAGE scan across all sites (Table A1). Each scan was quality checked and manually edited for movement, ghosting artifacts, or other incidental findings and processed with FreeSurfer 7.1.0 (del Re et al., 2021). Values of gyrification were defined through LGI, which is the ratio of the inner to outer smoothed cortical surface. LGI values were extracted across both hemispheres for the cortical lobes (the frontal, parietal, temporal, occipital, and cingulate) as well as the 33 individual regions defined through FreeSurfer cortical parcellation procedures using the Desikan-Killany cortical atlas and the insula (Desikan et al., 2006). Estimated total intracranial volume was extracted from FreeSurfer.

2.3. Analysis

Statistical analysis was conducted in RStudio. Demographic information for HC and probands was analyzed using ANOVA and chi-squared tests of independence (Nanda et al., 2013); sex, age, site of data acquisition, and estimated total intracranial volume were determined to be covariates. Probands were assessed with the BACS battery, and all BACS sub-scores were pre-adjusted for age and sex, z-transformed, and winsorized to +/− 3.0 standard deviations (Keefe et al., 2004).

LGI values were winsorized to +/− 3.0 standard deviations from the mean across both hemispheres for the 4 lobes, the cingulate, and all 34 individual regions. The average total cortical LGI for each subject was calculated to represent the gyrification of the whole brain, weighted by surface area. The effects of the covariates, excluding age, were taken into account by adjusting the whole brain LGI with a linear regression. The resulting total cortical LGI was analyzed in MATLAB R2022 to quantify the impact of age on gyrification.

2.3.1. Mathematical Model

We fit an appropriate model to adult HC and proband data from age 18 to 60. Standard mathematical functions, including linear, polynomial, exponential, and logarithmic, were tested as models for the changes of total cortical LGI with respect to age. The Least Squares Method was used through applying the fminsearch function in MATLAB to minimize the sum of squared differences between the data points and the models, as shown in Table A2 (Lagarias et al., 1998). The function of best fit was found to be the logarithmic function, as defined in [Eq.1]:

$$LGI=a+b*log(age+c)$$ [Eq.1]

In equation 1, a is an indicator of LGI levels independent of age, b is the decrease rate of LGI over age, and c is a translational term for age (Cao et al., 2017). The initial value of a used for all diagnosis group fitted gyrification curves was 2.994, because this value is the mean of the total cortical LGI in HC participants. The initial value of b was set to −1 to reflect LGI decreasing with age. To compare fitted gyrification curves across all samples without changing the shape or levels of the curves (Cao et al., 2017), the level of c was set to 0.003, according to the fitting for the HC sample. The logarithmic function, as defined in equation 1, was fit to data from each diagnosis group.

To determine if there is a significant difference between the fitted LGI curves for each diagnosis group, we estimated the distributions of fitting parameters a and b utilizing a resampling technique (Cao et al., 2017). Each diagnosis group was stratified into two age blocks split by the median age of the dataset: (i) probands between 18 and 36 years of age and (ii) probands between 37 and 60 years of age. Within each age block, a sample of 50% of the subjects was randomly selected without repetition. The two age block samples (i) and (ii) were then combined into one, randomly selected diagnosis group sample. Equation 1 was fit to the four randomly selected diagnosis group samples to find the best fit parameters a and b. This re-sampling procedure was repeated 10,000 times for each diagnosis group, and the fittings were performed each time—one for each diagnosis group sample. Hence, we extracted 10,000 sets of best fit parameters a and b for each diagnosis group.

Further pairwise comparisons were performed using the Tukey’s Honestly Significant Difference test by applying the multcompare function in MATLAB to determine which diagnosis group sample mean, derived from the best fit parameter distributions, was statistically significantly different.

2.3.2. Statistical Analysis

To investigate LGI differences between diagnosis groups—for probands 18 to 60, we used a two-step approach to run lobe and sub-region analyses on independent samples to prevent circular analysis (Nanda et al., 2013). First, a selection step was performed where ⅓ of the participants were randomly assigned into a sample used for analysis of the 4 lobes (frontal, parietal, temporal, occipital) as well as the cingulate in both hemispheres. Contrasts were conducted comparing LGI values bilaterally for the five regions across diagnosis group categorization. When a region in the selection step exhibited a trending difference in LGI (p<0.1), it was retained for the secondary step. In the secondary step, the remaining ⅔ of the participants after random selection were assigned to a sample used for analysis across both hemispheres for 34 individual regions. Contrasts between diagnosis groups were run for individual sub-regions that composed the region of trending difference identified through the selection step. All contrasts between diagnosis groups for the selection and secondary steps were conducted using a general linear model. Significance was set at p < 0.05 after false discovery rate (FDR) correction. The R-package ggseg 1.6.5 was used to visualize results (Mowinckel and Vidal-Piñeiro, 2020).

To probe if the comparatively higher and lower LGI curves for SZ compared to HC in the mathematical model were significant, pairwise general linear hypothesis tests (GLHT) were conducted comparing total cortical LGI between HC and SZ probands between the ages of 18 to 32 and the ages of 33 to 60. In regions where probands’ LGI differed from controls, LGI values were tested for correlations with BACS scores, FDR adjusting for the total number of correlations.

3. Results

3.1. Sample

The probands’ demographic information is summarized in Table 1. A total of 440 SZ, 350 SAD, 343 BP, and 727 HC individuals were included in our analysis.

Table 1.

Sample Demographics

	HC	SZ	SAD	BP	χ2 value / F-value	P-value
N	727	440	350	343
Age Mean ± SD	36.25 ± 12.24	36.60 ± 12.16	38.36 ± 11.90	36.26 ± 12.06	2.68	0.046
Sex (M/F)	313/414	281/159	160/190	131/212	65.36	< 0.001
Estimated Total Intracranial Volume Mean (SD)	1450979.81 (191654.56)	1463627.03 (203159.00)	1418015.03 (197562.32)	1444898.77 (203711.38)	3.65	0.012
Site of Data Acquisition					259.21	< 0.001
Baltimore	51	66	24	27
Boston	91	43	29	27
Chicago	161	97	74	124
Dallas	137	67	76	62
Detroit	40	34	5	25
Georgia	97	35	36	13
Hartford	150	98	106	65

3.2. Group Contrasts by DSM-IV Diagnosis

3.2.1. Lobe analysis.

We investigated gyrification in the frontal, occipital, temporal, parietal, and cingulate lobes in both hemispheres for analysis of DSM-IV groups. Significant widespread hypogyria was found in probands with psychotic disorders compared to HC (Figure 1). Significant hypogyria (p<0.01; d= −0.3 to −0.48) was observed across both hemispheres for all 4 lobes and the cingulate in SZ compared to HC. SAD probands had significant hypogyria in the left frontal (p<0.05, d= −0.3), right frontal (p<0.05, d= −0.28), left parietal (p<0.05, d= −0.31), left temporal (p<0.05, d= −0.29), right temporal (p<0.05, d= −0.32), right occipital (p<0.05, d= −0.29), left cingulate (p<0.01, d= −0.38), and right cingulate (p<0.001, d= −0.43) compared to HC. BP probands had hypogyria compared to HC only in the right temporal (p<0.05; d= −0.32). Significantly higher gyrification in BP compared to SZ was observed in the left frontal lobe (p<0.01; d= 0.41) and right frontal lobe (p<0.05, d= 0.27).

Figure 1.

Descriptive and comparative statistics of LGI in regions of significant and non-significant differences in gyrification between diagnosis groups after GLHT. Numbers represent Cohen’s d values and cells are color coded to the Cohen’s d scale [0.6, −0.6] as indicated to the right of the table. *** corresponds to p value < 0.001, ** corresponds to p value < 0.01, and * corresponds to p value < 0.05.

3.2.2. Desikan-Killany cortical atlas sub-regions

We further investigated gyrification bilaterally across 34 individual brain sub-regions. In SZ-HC comparisons, a total of 68 individual sub-regions were studied because the selection step identified bilateral hypogyria in all 4 lobes and the cingulate; hypogyria was found in 53 individual regions (p<0.05; d= −0.22 to −0.54) for SZ-HC comparisons after FDR correction (Figure 2; Table A3). Significant hypogyria was observed across 40 individual regions (p<0.05, d=−0.24 to −0.44) for SAD-HC comparisons (Figure 2; Table A4). Significant hypogyria was observed in BP compared to HC in 2 individual regions (Figure 2; Table A5): the right superior temporal (p=0.026, d= −0.36) and right middle temporal (p=0.021, d= −0.36).

Figure 2.

LGI alterations in regional contrasts between BP and HC; SAD and HC; and SZ and HC. Colored regions indicate regions of significant differences in LGI, and the gradient corresponds to effect sizes.

No significant differences in gyrification were found across sub-regions for any of the BP-SZ, SZ-SAD, or SAD-BP comparisons.

3.3. Relationship between LGI and BACS scores

A statistically significant yet weak relationship was found for SZ probands between BACS Verbal Memory Scores and LGI in the right frontal (p=0.03, r= 0.26) and right cingulate (p=0.01, r= 0.29) regions after FDR correction. (Table 2).

Table 2.

Correlations for BACS Verbal Memory Scores and LGI regions in SZ probands.

Group	BACS Test	ROI	r	P-value
SZ	BACS Verbal Memory	Right Frontal	0.26	0.03
		Right Cingulate	0.29	0.01

3.4. Fitted Gyrification Curves

3.4.1. DSM-IV Diagnosis Groups

Fitted LGI curves for diagnosis groups were estimated individually using Equation 1, from age 18 to 60. In Figure 3, the LGI curves to probands with psychotic disorders diverge from that to HC. Across all ages from 18 to 60, the LGI curve of BP probands is lower than HC (Figure 3; Table A6). SAD probands’ curve overlaps with HC at early ages until about the age of 25, after which the SAD curve is lower and diverging from the HC curve (Figure 3; Table A6). SZ probands’ curve show higher total cortical LGI than HC at young ages (between 18 to 32), and lower total cortical LGI after approximately the age of 32 compared to the HC curve (Figure 3; Table A6).

Figure 3.

Comparative plot of total cortical LGI vs. Age in years for HC (dark blue), SZ (green), SAD (red), and BP (light blue).

To demonstrate that there is a statistically significant difference in the fitted LGI curves between diagnosis groups, we performed the resampling technique described in the methods to show that the best fit coefficients a and b for each of the fitted curves are significantly different. We found that fitting parameter a was lower in HC than in SZ (p<0.001) and SAD (p<0.001), and higher in HC than in BP (p<0.001) (Table A7). Fitting parameter b, which indicates the rate of LGI decline, was found to be higher (less negative) in HC than in SZ (p<0.001) and SAD (p<0.001), and lower (more negative) in HC than in BP (p<0.001) (Table A7).

3.4.2. Group Contrasts between SZ and HC probands before and after the age of 32.

The results from 3.4.1, which reveal hypergyria in SZ probands before the age of 32 and hypogyria in SZ after the age of 32 compared to HC, are reinforced after pairwise GLHTs. Pairwise tests comparing total cortical LGI in SZ and HC probands between the ages of 18 to 32 show statistically significant hypergyria in SZ (d= 0.31, p<0.001); pairwise tests in probands between the ages of 33 to 60 show significant hypogyria (d= −0.36, p<0.001) in SZ compared to HC (Table 3).

Table 3.

Results of total cortical LGI comparison between diagnosis groups SZ and HC after pairwise GLHT comparison for two separate age groups: 18 – 32 and 33 – 60.

Diagnosis Groups Contrasted	Region Studied	Age	T statistic	Adjusted P-value	d	CI lower	CI upper
SZ-HC	Total Cortical LGI	18 – 32	3.33	<0.001	0.31	0.125	0.487
		33 – 60	−4.36	<0.001	−0.36	−0.517	−0.194

4. Discussion

We employed a large dataset of psychosis probands and non-psychosis controls to investigate differences in gyrification across psychotic disorders and characterized the relationship between total cortical LGI and age of participants.

Pairwise contrasts of lobes and subregions for SZ to HC probands revealed widespread hypogyria bilaterally for the frontal, parietal, temporal, occipital, and cingulate. The finding of widespread hypogyria in SZ is corroborated with results from previous studies (Madre et al., 2020; Palaniyappan and Liddle, 2012; Spalthoff et al., 2018). Analysis of the composite subregions’ gyrification finds 53 significant subregions of hypogyria in SZ compared to HC.

We found a positive correlation between BACS Verbal Memory Test scores and gyrification within the right frontal lobe for SZ. The link between SZ gyrification and cognitive results is intriguing as it relates to previous findings within the field. One study found the severity of positive psychotic symptoms, such as auditory verbal hallucinations, is correlated with right frontal LGI in probands with SZ (Sasabayashi et al., 2021). Another study in HC linked better verbal working memory to increased gyrification in the bilateral superior and medial frontal cortex (Gautam et al., 2015).

We found a correlation between BACS Verbal Memory Test scores and gyrification within the right cingulate for SZ probands. The cingulate interacts with many brain networks and participates in the formation and processing of emotions, in learning and memorizing, and is responsible for executive functions (Zakharova et al., 2021). In SZ probands, the anterior cingulate cortex—which is involved in cognitive tasks like verbal selection and workingmemory—has been shown to have abnormal activity during cognitive task performance (Bush et al., 2000; Adams and David, 2007). Hence, our finding of hypogyria in the right cingulate in SZ probands may be relevant to the underlying cognitive dysfunction and abnormal activity associated with SZ.

We found hypogyria in the right temporal lobe for BP compared to HC. Analysis of the composite subregions’ LGI reveals statistically significant hypogyria in the right superior temporal and middle temporal. These results are supported by a previous study that found significant hypogyria in BP probands in the superior temporal cortex (Choi et al., 2020). Comparing BP to SZ probands reveals statistically significantly higher LGI in the left frontal lobe of BP compared to SZ. Though few, there have been previous studies comparing gyrification within BP and SZ probands in the frontal lobe, and statistically significant higher LGI in BP compared to SZ has been localized to the inferior frontal gyrus and lateral orbitofrontal (Miola et al., 2022). Considering some overlap in SZ and psychotic BP morphometric features (Ivleva et al., 2010), these findings of higher LGI within the left frontal lobe for BP suggest a differential pattern of gyrification between the two DSM-IV categorizations.

Our study is the first to report hypogyria in SAD in the left and right cingulate; left and right temporal; left and right frontal; right occipital; and left parietal. Furthermore, we are the first to report hypogyria across 40 individual subregions in SAD compared to HC. Interestingly, all three psychotic disorders exhibit hypogyria in the right superior temporal compared to HC. Our findings of bilateral hypogyria within the cingulate for both SAD and SZ and hypogyria in the temporal regions for SAD bilaterally, SZ bilaterally, and BP in the right hemisphere suggest that the disorders may be characterized by abnormal gyrification within the cingulate and temporal regions.

Our results show that the fitted LGI curves for HC and DSM-IV categorized probands follow a logarithmic function with respect to age, confirming previous research in HC (Lamballais et al., 2020; Madan and Kensinger, 2016). We find that the age-related gyrification curve for BP probands is lower than the HC curve. Heterogeneous findings of both hypogyria and hypergyria have been previously identified for BP probands in the literature (Sasabayashi et al., 2021). Intriguingly, studies which do suggest hypergyria in cortical regions for BP have shown the simultaneous presence of hypergyria and hypogyria in BP probands, suggesting that mixed changes in gyrification or altered lateralization of gyrification across the brain may be present in BP (Nenadic et al., 2015; Palaniyappan and Liddle, 2014). As our curve uses a total cortical LGI estimation, such regional nuance in gyrification are not captured.

Of interest for SZ, our model’s novelty is that it captures both hypergyria, for SZ probands before age 32, and hypogyria, for SZ probands after 32, compared to controls, which supports our hypothesis. Previous studies of gyrification in young adult and adolescent probands with SZ (before the age of 30) have reported hypergyria compared to HC (Sasabayashi et al., 2021). Meanwhile, studies in adult SZ probands with a mean age roughly greater than 30 have revealed hypogyria compared to HC (Sasabayashi et al., 2021). Until now, no study has captured both hypergyria and hypogyria simultaneously in one model. Hypogyria in SZ after age 32 is consistent with the hypothesis of an accelerated gyrification decline across adulthood, potentially indicative of accelerated aging in SZ (Pham et al., 2021; Yunzhi et al., 2022). Of note, until the present study, accelerated aging has been thought to explain reductions in cortical thickness (del Re et al., 2021; Storsve et al., 2014; Shahab et al., 2018; Thormodsen et al., 2013; del Re and Keshavan, 2023), surface area (Storsve et al., 2014), and volume (Storsve et al., 2014; Schnack et al., 2016) but has not been implicated in age related changes in gyrification. On the other hand, the observed increases in LGI in SZ before age 32 may be related to different pathophysiological processes early in the illness (Cetin-Karayumak et al., 2023; Di Biase et al., 2021). For example, there is evidence that hippocampal volumes may increase early in the illness, followed by later declines (Li et al., 2018). Furthermore, these results suggest that present discordant results in the gyrification literature in SZ may be attributed to both disease related pathophysiological processes early in the illness and age-related degenerative processes.

There are some limitations to our analyses. The fitted LGI curves were developed using a large sample of cross-sectional data spanning the ages of 18 to 60; hence, longitudinal data may be necessary to compare differences in gyrification with respect to age. Nonetheless, the current analysis evaluating the relationship between age and gyrification is promising. The present study suggests that gyrification of the human brain in vivo decreases non-linearly throughout adulthood. Though our study covers a large age-range, our model doesn’t account for gyrification during infancy, childhood, and early adolescence. Despite these limitations, our study is novel and sheds new light on gyrification across psychosis spectrum disorders through its findings of statistically significant hypogyria and differential fitted LGI curves.

5. Conclusion

In this cross-sectional study using a large, transdiagnostic dataset, we have observed hypogyria in probands with psychotic disorders compared to controls and differences in gyrification between psychotic disorder probands; taken together, these results suggest differential patterns of gyrification across psychotic disorders. We created a mathematical model of gyrification with respect to age for psychotic disorder probands and controls, which uncovered age-related differences in gyrification. Furthermore, the novel findings reveal hypergyria in early adulthood and later hypogyria after the age of 32 in probands with SZ compared to controls. While the model is cross-sectional, it is the first to address discordant results within the literature on gyrification in psychotic disorders by demonstrating age-related changes.

Appendix

Table A1.

Table of Scanning Parameters.

	Boston 1	Boston 2	Boston 3	Georgia	Chicago 1	Chicago 2	Dallas 1	Dallas 2	Detroit	Baltimore	Hartford 1	Hartford 2	Hartford 3
Manufacturer/model	Signa Excite	Signa HDxt	Signa Discovery MR750	Signa HDxt	Signa HDx	Philips Achieva	Philips Achieva		Siemens Triotim	Siemens Triotim,	Siemens Triotim	Siemens Skyra
Scanner location	Beth Israel Deaconess Medical Center, Boston MA			University of Georgia, Athens, GA	University of Illinois, Chicago IL	University of Chicago, Chicago, IL	UT Southwestern Medical Center, Dallas. TX		Harper University Hospital, Detroit, Ml	John Hopkins Hospital, Baltimore MD	Olin Neuropsychiatry Research Center, Hartford CT
Field Strength (T)	3	3	3	3	3	3	3	3	3	3	3	3	3
Slice Thickness (mm)	1.2	1.2	1.2	1.2	1.2	1.2	1.2	1.2	1.2	1.2	1.2	1.2	1.2
Repetition Time (ms)	6.744	6.988	7.38	6.968	6.988	6.788	6.809	6.8	2300	2300	2300	2300	2300
Echo Time (ms)	2.996	2.848	3.052	2.832	2.996	3.10	3.13	3.10	2.89	2.91	2.95	2.95	2.95
Inversion Time (ms)	1100	650	400	400	1100	850	850	850	900	900	900	900	900
Flip Angle	8	8	11	11	8	9	8	9	9	9	9	9	9
In-plane resolution matrix (Z * Y)	256 × 256	256 × 256	256 × 256	256 × 256	256 × 256	256 × 256	256 × 256	256 × 256	256 × 240	256 × 240	256 × 240	256 × 240	256 × 240
Parallel?	No	No	Yes	Yes	No	Yes	No	Yes	No	No	No	No	Yes

Note: T, Tesla; mm, millimeter; ms, millisecond

Table A2.

Comparisons between different fitting functions. Functions fitted to total cortical LGI for HC ages 18 – 60.

Mathematical Function	a 0	b 0	c 0	d 0	Sum of Squared Differences
a + bX	3.004	−3.302×10−4	-	-	0.036
a + bXc	3.805	−0.816	−6.829×10−4	-	0.037
a + b(X + c)d	2.994	−1	1.000×10−3	1.000×10−3	0.036
a + bX+c	2.991	−0.525	8.243×10−4	-	0.036
a + b * log(X + c)	3.040	−0.014	0.004	-	0.035

Table A3.

Table of SZ-HC comparisons reporting p-values and effect sizes for significant subregions of LGI differences in SZ-HC comparisons.

SZ-HC Subregion Comparisons
Lobe	Region	Effect Size	P-value
Frontal	Left Superior Frontal	−0.41	<0.001
	Left Rostral Middle Frontal	−0.48	<0.001
	Left Pars Opercularis	−0.34	<0.001
	Left Pars Triangularis	−0.32	<0.001
	Left Pars Orbitalis	−0.26	0.015
	Left Lateral Orbitofrontal	−0.33	<0.001
	Left Medial Orbitofrontal	−0.28	<0.001
	Left Precentral	−0.37	<0.001
	Left Frontal Pole	−0.35	<0.001
	Right Superior Frontal	−0.42	<0.001
	Right Rostral Middle Frontal	−0.54	<0.001
	Right Pars Opercularis	−0.47	<0.001
	Right Pars Triangularis	−0.38	<0.001
	Right Pars Orbitalis	−0.32	<0.001
	Right Lateral Orbitofrontal	−0.41	<0.001
	Right Medial Orbitofrontal	−0.43	<0.001
	Right Precentral	−0.44	<0.001
	Right Frontal Pole	−0.45	<0.001
Parietal	Left Supramarginal	−0.28	<0.001
	Left Postcentral	−0.35	<0.001
	Left Precuneus	−0.23	0.027
	Right Inferior Parietal	−0.28	<0.001
	Right Supramarginal	−0.28	<0.001
	Right Postcentral	−0.35	<0.001
	Right Precuneus	−0.31	<0.001
Temproal	Left Superior Temporal	−0.39	<0.001
	Left Middle Temporal	−0.33	<0.001
	Left Fusiform	−0.33	<0.001
	Left Transverse Temporal	−0.37	<0.001
	Left Entorhinal	−0.30	<0.001
	Left Temporal Pole	−0.26	0.013
	Left Parahippocampal	−0.34	<0.001
	Right Superior Temporal	−0.39	<0.001
	Right Middle Temporal	−0.32	<0.001
	Right Fusiform	−0.27	0.01
	Right Transverse Temporal	−0.36	<0.001
	Right Entorhinal	−0.38	<0.001
	Right Temporal Pole	−0.30	<0.001
	Right Parahippocampal	−0.29	<0.001
Occipital	Left Lateral Occipital	−0.24	0.021
	Left Lingual	−0.37	<0.001
	Left Pericalcarine	−0.26	0.013
	Right Lingual	−0.28	<0.001
	Right Cuneus	−0.35	<0.001
	Right Pericalcarine	−0.33	<0.001
Cingulate	Left Rostral Anterior Cingulate	−0.40	<0.001
	Left Caudal Anterior Cingulate	0.38	<0.001
	Left Isthmus Cingulate	−0.26	0.013
	Right Rostral Anterior Cingulate	−0.45	<0.001
	Right Caudal Anterior Cingulate	−0.38	<0.001
	Right Posterior Cingulate	−0.22	0.034
	Right Isthmus Cingulate	−0.26	0.015

Table A4.

Table of SAD-HC comparisons reporting significant subregions and corresponding p-values and effect sizes for LGI differences.

SAD-HC Subregion Comparisons
Lobe	Region	Effect Size	P-value
Frontal	Left Superior Frontal	−0.33	0.013
	Left Rostral Middle Frontal	−0.31	0.017
	Left Pars Opercularis	−0.29	0.023
	Left Pars Triangularis	−0.27	0.028
	Left Pars Orbitalis	−0.29	0.023
	Left Lateral Orbitofrontal	−0.29	0.023
	Left Precentral	−0.34	0.013
	Left Frontal Pole	−0.25	0.037
	Right Superior Frontal	−0.30	0.022
	Right Rostral Middle Frontal	−0.34	0.013
	Right Pars Opercularis	−0.33	0.013
	Right Pars Triangularis	−0.37	0.013
	Right Pars Orbitalis	−0.36	0.013
	Right Lateral Orbitofrontal	−0.44	<0.001
	Right Medial Orbitofrontal	−0.32	0.015
	Right Precentral	−0.35	0.013
	Right Paracentral	−0.35	0.013
	Right Frontal Pole	−0.33	0.013
Parietal	Left Supramarginal	−0.25	0.037
	Left Postcentral	−0.29	0.023
	Right Supramarginal	−0.27	0.028
	Right Postcentral	−0.31	0.017
	Right Precuneus	−0.27	0.03
Temporal	Left Superior Temporal	−0.29	0.024
	Left Fusiform	−0.28	0.024
	Left Transverse Temporal	−0.35	0.013
	Left Parahippocampal	−0.28	0.024
	Right Superior Temporal	−0.33	0.013
	Right Fusiform	−0.30	0.022
	Right Transverse Temporal	−0.24	0.047
	Right Entorhinal	−0.27	0.028
	Right Parahippocampal	−0.34	0.013
Occipital	Right Pericalcarine	−0.27	0.028
Cingulate	Left Caudal Anterior Cingulate	−0.33	0.013
	Left Posterior Cingulate	−0.26	0.034
	Left Isthmus Cingulate	−0.25	0.037
	Right Rostral Anterior Cingulate	−0.28	0.025
	Right Caudal Anterior Cingulate	−0.29	0.023
	Right Posterior Cingulate	−0.41	<0.001
	Right Isthmus Cingulate	−0.32	0.015

Table A5.

Table of BP-HC comparisons reporting significant subregions and corresponding p-values and effect sizes for LGI differences.

BP-HC Subregion Comparisons
Lobe	Region	Effect Size	P-value
Temporal	Right Superior Temporal	−0.33	0.026
	Right Middle Temporal	−0.36	0.021

Table A6.

Best fit parameters a, b, and c used for Figure 3 plots.

	Diagnosis Group
	HC	SZ	SAD	BP
a	3.041	3.373	3.082	2.968
b	−0.014	−0.109	−0.029	0.002
c	0.003	0.003	0.003	0.003

Table A7.

Multiple comparison test between means of best fit parameter a and parameter b’s distributions for Diagnosis Groups Figure 3.

Group	Control Group	Parameter a		Parameter b
		Difference	P-value	Difference	P-value
HC	SZ	−0.331	< 0.001	0.095	< 0.001
HC	SAD	−0.040	< 0.001	0.015	< 0.001
HC	BP	0.074	< 0.001	−0.016	< 0.001
SZ	SAD	0.291	< 0.001	−0.080	< 0.001
SZ	BP	0.405	< 0.001	−0.112	< 0.001
SAD	BP	0.114	< 0.001	−0.032	< 0.001