Schizophrenia Research 159 (2014) 435–440
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
A diffusion tensor imaging family study of the fornix in schizophrenia Vina M. Goghari a,⁎, Thibo Billiet b, Stefan Sunaert b, Louise Emsell b a b
Clinical Neuroscience of Schizophrenia (CNS) Laboratory, Department of Psychology, Hotchkiss Brain Institute, University of Calgary, 2500 University Drive, NW Calgary, AB T2N 1N4, Canada Translational MRI, Department of Imaging and Pathology, KU Leuven and Radiology, University Hospitals Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 2 August 2014 Received in revised form 18 September 2014 Accepted 22 September 2014 Available online 12 October 2014 Keywords: Psychosis Tractography Fornix Family study Hippocampus White matter
a b s t r a c t Diffusion tensor imaging (DTI) studies suggest abnormalities in the white matter microstructure of the fornix in schizophrenia patients. Research evaluating schizophrenia patient and relatives also suggests that the white matter microstructure of the fornix is heritable. However, previous studies have been hindered by limited DTI methodology. Therefore, the goal of this study was to assess whether fornix abnormalities were related to the genetic liability for schizophrenia using the novel methodological approach of assessing multiple metrics of along-tract measurements, in addition to whole-tract means. Twenty-five schizophrenia patients, 24 adult non-psychotic first-degree biological relatives, and 27 community controls underwent neuroimaging. No group differences were found for any of the DTI metrics using the classical whole-tract measures of the fornix. Along-tract analysis detected local increases in fractional anisotropy (FA) in the right fimbria of the fornix for relatives compared to patients and controls corrected for false discovery rate. No significant associations were found between symptoms, global functioning, or IQ and whole-tract FA means in schizophrenia patients or relatives. Increased FA in nonpsychotic relatives could represent a compensatory mechanism to guard against psychosis or an abnormality associated with the genetic liability for the disorder. These findings underscore the importance of obtaining alongtract measurements, in addition to whole-tract measurements to fully understand white matter abnormalities in schizophrenia. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Diffusion tensor imaging (DTI) metrics differentiate between schizophrenia patients and controls, and furthermore, appear to be useful vulnerability markers for schizophrenia (e.g., Camchong et al., 2009; Clark et al., 2011; Knochel et al., 2012; Skudlarski et al., 2013). Additionally, these measures are associated with symptoms of the disorder in both patients and relatives (Knochel et al., 2012). Although most white matter (WM) tracts have been investigated, most DTI studies of relatives have not assessed the fornix. The fornix is a tract of key interest in schizophrenia, given it connects the hippocampus to the hypothalamus. Several studies in schizophrenia patients have found WM abnormalities in the fornix using voxel-based (e.g., Guo et al., 2012) and tract-based DTI analyses (e.g., Fitzsimmons et al., 2014). While tractography itself is not a quantitative technique (it is a way of virtually reconstructing WM fiber-like pathways), it is often used synonymously with the method used to extract mean quantitative values of tracts generated from tractography (Jbabdi and Johansen-Berg, 2011). However, it is known that DTI metrics vary along the length of tracts; therefore, taking a simple mean of all these ⁎ Corresponding author at: Department of Psychology, University of Calgary, 2500 University Drive, NW Calgary, AB T2N 1N4, Canada. Tel.: +1 403 210 7344; fax: +1 403 282 8249. E-mail address: [email protected] (V.M. Goghari).
http://dx.doi.org/10.1016/j.schres.2014.09.037 0920-9964/© 2014 Elsevier B.V. All rights reserved.
values, as in classical tractography analysis, may be insufficient for capturing more subtle local differences (Colby et al., 2012). A more sophisticated approach that accounts for this variation is along-tract analysis, whereby focal differences in parameters manifest as local offsets between along-tract profiles (i.e., the plot of metric value at predetermined points along a tract) and whole-tract differences manifest as a global off-set between tract profiles. Abdul-Rahman et al. (2011) used along tract-based analysis and found specific loci for fractional anisotropy (FA) reductions and axial diffusivity (AD) and radial diffusivity (RD) increases in the left fornix. Furthermore, this study found decreased FA in the specific part of the left fornix that was related to increased psychotic symptoms in patients (Abdul-Rahman et al., 2011). Along-tract methods have not been used in any family studies of schizophrenia patients. Using traditional methods, two large scale studies, including over 100 relatives and controls each, did not find FA differences in the fornix in relatives versus controls (Boos et al., 2013; Skudlarski et al., 2013). However, one study found that FA values of the fornix were heritable in schizophrenia patients and relatives (Skudlarski et al., 2013), suggesting that further investigation of the fornix with more sophisticated DTI methods, in relatives, is warranted. Most previous studies investigating WM changes in schizophrenia have used voxel-based analysis, typically of a highly restricted WM skeleton (e.g., tract based spatial statistics), which is not an optimal approach for investigating a narrow tract such as the fornix (Bach et al.,
436
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
2014). Other studies have used whole-tract mean values from tractography; thereby, losing anatomical specificity. The goal of this investigation therefore, was to use advanced DTI analysis to examine along-tract microstructural differences, in addition to whole-tract mean values, including measures of FA, mean diffusivity (MD), RD, and AD, in two subdivisions of the fornix in a family study of schizophrenia. We expected significant differences between all groups, with greater along-tract differences than whole-tract mean differences. We hypothesized that the largest differences would be found between patients and controls with non-psychotic relatives being intermediate. In terms of the individual metrics, we expected reduced FA and increased MD, RD, and AD in patients and relatives compared to controls in both parts of the fornix. 2. Materials and methods
families in which two relatives participated and one family in which three relatives participated. Community controls were recruited through flyers and advertisements around the community. The University of Calgary Ethics Board approved the protocol and informed written consent was obtained. Inclusion criteria for all participants included: (1) age 18–65; (2) minimum IQ of 70; (3) no current diagnosis of drug/alcohol dependence/abuse; (4) no history of head injury or being unconscious for more than 20 min; (5) no history of electroconvulsive therapy; and (6) no history of stroke or other neurological condition. Further criteria for inclusion of first-degree relatives were no lifetime diagnosis of a psychotic disorder or bipolar disorder, or history of anti-psychotic medication use. Further criteria for inclusion of community controls were no personal or family history of a psychotic disorder or bipolar disorder, or personal use of an anti-psychotic medication.
2.1. Participants 2.2. Diagnosis and assessment A total of 76 individuals participated: 25 schizophrenia or schizoaffective patients (hereafter referred to as schizophrenia patients), 24 adult non-psychotic first-degree biological relatives, and 27 community controls. Demographic characteristics are shown in Table 1. Schizophrenia patients were recruited through outpatient clinics and through community support programs in Calgary, Canada. Research staff identified first-degree biological relatives by completing a pedigree with the proband. Not all probands met recruitment criteria for the study; however, to enhance the sample, all first-degree biological relatives of schizophrenia patients that met recruitment criteria were included. Seven relatives were related to probands in this sample and there were two
Participants were interviewed using the Structured Clinical Interview for DSM-IV Axis I Disorders. The Structured Interview for Schizotypy, with supplemental questions, was used to measure Axis II Cluster A disorders in relatives and controls (Kendler et al., 1989). Diagnoses were assigned according to DSM-IV-TR criteria via case conferences. No relatives or controls met criteria for a Cluster A disorder. Symptoms and functioning were measured using the Positive and Negative Syndrome Scale (PANSS; Kay et al., 1987), Social Functioning Scale (Birchwood et al., 1990), the Global Assessment of Functioning (GAF) Scale (APA, 1994). Lastly, the vocabulary and matrix reasoning subtests of the Wechsler
Table 1 Participant characteristics: demographics, IQ measures, symptoms, functioning, and medication usage.
N Age Gender (% female) Born in Canada (%) Education (years completed) Annual income (%)1 $0–$30,000 $30,000–$50,000 $50,000–$95,000 $95,000+ Maternal education (years completed) Paternal education (years completed) Matrix reasoning raw score Vocabulary raw score Handedness (% right handed) PANSS negative: range PANSS positive: range PANSS general: range Global assessment of functioning: range Social functioning scale: range Axis I (% with any lifetime diagnosis) Relative status — parent:sibling:offspring Anti-psychotic (atypical, typical, both; % on) Anti-depressants (% on) Mood stabilizer (% on) Anti-anxiety (% on) Anti-parkinson (% on) Other psychiatric (% on)
Schizophrenia
Relative
Control
25 41.3 (10.8) 48 92 14.2 (3.0)a
24 40.2 (15.0) 58 88 16.4 (2.6)
27 40.7 (11.1) 52 81 15.3 (2.4)
64 20 12 4 13.7 (2.8) 14.0 (3.0) 26.3 (2.8) 57.1 (5.9)a 92 11.9 (3.6): 7–19 14.4 (5.3): 7–24 26.8 (6.5): 16–39 53.4 (14.9): 30–83a,c
4 17 54 25 13.1 (3.8) 12.6 (4.0) 27.6 (3.1) 62.2 (5.8) 79 7.8 (1.1): 7–11b 8.8 (1.9): 7–14b 20.5 (3.6): 16–29b 81.8 (5.7): 63–88
– N/A 96, 4, 4 40 12 8 4 8
332 9:12:3 0, 0, 0 8 0 4 0 4
8 19 39 35 13.5 (3.4) 13.7 (4.7) 27.0 (5.6) 59.1 (7.8) 96 7.2 (0.6): 7–10b 7.9 (1.3): 7–11b 18.1 (37): 16–33b 84.7 (5.1): 73–95 793.5 (53.1): 701–883 263 N/A 0, 0, 0 8 0 0 0 0
Note. Mean and standard deviation presented where appropriate. The following notations were used for non-medication variables: a Less than relatives. b Less than patients. c Less than controls. 1 Overall chi-square demonstrated patients had more individuals in lower income brackets than controls or relatives. 2 Some relatives had more than one lifetime Axis I disorder: 1 with alcohol dependence, 4 with alcohol abuse, 1 with cannabis abuse, 5 with major depressive disorder, 1 with dysthymia, 2 with post-traumatic stress disorder, 1 with panic disorder without agoraphobia, and 1 with generalized anxiety disorder. 3 Some controls had more than one lifetime Axis I disorder: 1 with alcohol dependence, 1 with cocaine dependence, 3 with alcohol abuse, 1 with hallucinogen abuse, 1 with cannabis dependence, 3 with major depressive disorder, and 1 with social anxiety disorder.
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
(a)
Body
437
(b)
Fimbria
(i)
Fornix
FA
MD
(ii)
Le body
RD (iii)
Le fimbria
AD
Le
Right Controls
Le Relaves
Right
Schizophrenia
Fig. 1. Diffusion tensor imaging of the fornix in schizophrenia patients, non-psychotic relatives, and controls. (a) Changes in DTI parameters along the length of the fornix body and fimbria subdivisions between schizophrenia patients, non-psychotic relatives, and community controls. The x-axis describes the position along the tract, beginning to end at intervals of 1 mm. In the body divisions, it represents points 0 to 24, and in the left/right fimbria points 0 to 74/78. The along-tract analysis assessed both changes in the mean offset (i.e., tract mean) between the profiles (i.e. differences between the red, green and blue lines), and local (i.e., tract position) differences. The significant local difference in FA between relatives and patients and controls can be seen in the mid-fimbria. (b) The three-dimensional tract masks used to represent the whole fornix (i), colored according to the principal direction of diffusion, and to generate its subdivisions, the body (left hemisphere) (ii) and fimbria (left hemisphere) (iii), both overlaid on the population atlas color FA map. Abbreviations: FA: fractional anisotropy, MD: mean diffusivity, AD: axial diffusivity, and RD: radial diffusivity. FA is a fraction, without units. Units of MD, RD & AD = mm2/s.
Abbreviated Scale of Intelligence were used to estimate IQ (WASI; Wechsler, 1999).
2.3. Diffusion tensor imaging High-angular resolution diffusion (HARDI) MRI data were acquired (3 T GE) along 60 gradient directions with b-value 1300 s/mm2 and corrected for motion and distortion using ExploreDTI (Leemans et al., 2009). DTI parameters FA, MD, RD, and AD were estimated. A population-based template was constructed from the participant datasets, to which all diffusion tensor maps were registered (Van Hecke et al., 2008). This approach utilized non-affine registration and tensor reorientation, in order to correctly preserve the orientation information captured in the diffusion signal (Van Hecke et al., 2007). Parameter maps were obtained in atlas space. The fornix was isolated in template space using targeted deterministic tractography in ExploreDTI (Emsell et al., 2013). Seed-based tracking was initiated in the fornix body, proceeded with a step size of 1 mm and terminated in voxels with an FA of N 0.12. Spurious fibers not belonging to the fornix were removed by tract selection regions of interest. Subdivision of the whole tract into left/right body and left/right fimbria was performed to ensure consistent streamline lengths for along-tract sampling. The body and fimbria subdivisions were defined as the point at which the whole body diverged into two tracts (i.e., the fimbria). The anterior boundary of the body was defined as the point at which it divided into the two
fornix columns (Fig. 1b). In-house software was used to calculate parameter values at 1 mm intervals along the length of the tract masks for each subject in template space. For the left and right body, 24 points were assessed in each tract. For the left fimbria, 74 points were assessed, and for the right fimbria, 78 points. For the along-tract analysis, average parameter values were taken at each of these evenly spaced points along the reconstructed tractography streamlines comprising the four fornix segments (i.e., left/right body and left/right fimbria) in each participant separately. Classical whole-tract means were also calculated by dividing the total metric value at every point by the number of points per tract in each participant.
2.4. Statistical analyses Statistical analysis softwares used were SPSS v 21 and MATLAB version 2011b. Age and gender were used as covariates in all analyses, as there is evidence from other studies that DTI parameters may vary as a function of age (Lebel et al., 2012) and gender (Sacher et al., 2013). Greenhouse–Geisser correction is reported for analyses of covariance (ANCOVAs). Partial eta-squared effect sizes (η2) are also provided. For the classical whole-tract means, four 2 hemisphere (left, right) by 3 group ANCOVAs were used to assess for the effect of group on FA, MD, RD, and AD separately for the body and fimbria of the fornix. For comparison to previous literature and inclusion in meta-analyses, Cohen's d effect sizes are presented for all group comparisons of the whole-
438
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
Table 2 Raw classical whole-tract mean values and absolute Cohen's d effect size values for group comparisons for DTI metrics of body and fimbria of the fornix.
Body FA left MD left RD left AD left FA right MD right RD right AD right Fimbria FA left MD left RD left AD left FA right MD right RD right AD right
Schizophrenia
Relative
Control
Schizophrenia vs control Cohen's d
Relative vs control Cohen's d
Schizophrenia vs relative Cohen's d
0.43 (0.06) 0.0018 (0.0002) 0.0015 (0.0002) 0.0023 (0.0002) 0.43 (0.06) 0.0018 (0.0002) 0.0015 (0.0002) 0.0022 (0.0001)
0.45 (0.03) 0.0017 (0.0001) 0.0014 (0.0001) 0.0022 (0.0001) 0.43 (0.05) 0.0017 (0.0001) 0.0015 (0.0001) 0.0022 (0.0001)
0.43 (0.04) 0.0017 (0.0001) 0.0015 (0.0001) 0.0022 (0.0001) 0.43 (0.04) 0.0017 (0.0001) 0.0015 (0.0001) 0.0022 (0.0001)
0.05 0.16 0.13 0.20 0.01 0.18 0.15 0.25
0.53 0.40 0.44 0.31 0.10 0.22 0.20 0.24
0.33 0.52 0.52 0.50 0.09 0.38 0.33 0.45
0.39 (0.03) 0.0014 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001) 0.41 (0.03) 0.0013 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001)
0.40 (0.04) 0.0013 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001) 0.41 (0.04) 0.0013 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001)
0.38 (0.04) 0.0014 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001) 0.40 (0.03) 0.0013 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001)
0.27 0.01 0.04 0.11 0.07 0.13 0.10 0.19
0.48 0.40 0.04 0.32 0.30 0.16 0.18 0.11
0.25 0.43 0.43 0.43 0.23 0.28 0.27 0.29
Mean and standard deviation. Cohen's d was calculated using as many decimal places as available. FA = fractional anisotropy, MD = mean diffusivity, RD = radial diffusivity, AD = axial diffusivity. Units of MD, RD and AD = mm2/s.
tract raw values in Table 2. Follow-up testing was only conducted if the 3 group ANCOVA was significant to control for multiple comparisons. Along-tract analysis was used to detect both local differences in DTI measures and global offset of the tract profiles. Whereas the classical approach assesses the difference between 3 points (i.e. the tract mean of group 1, tract mean group 2, tract mean group 3), in the along-tract approach, there is a repeated measures term of ‘position’ as a result of variance in metrics at each point along the tract. The along-tract approach accounts for this variance by investigating the offset between 3 spatially varying curves (Colby et al., 2012). This requires a different statistical model with many more terms (i.e., tract points). Twelve ANCOVAs (4 metrics × 3 two-group contrasts) with group, position along the tract, and group by position interaction were used to assess for the effect of group on FA, MD, RD, and AD separately for the body and fimbria of the fornix. As this was a novel method, we a priori directly compared each two groups (patients vs. controls, relatives vs. controls, relatives vs. patients). False discovery rate (FDR) correction was applied to all along-tract analyses as a whole to control for multiple comparisons. Lastly, partial correlations were conducted to assess the relationship between FA and symptoms, global functioning, and IQ.
3. Results
Groups differed in their positive, negative, and general PANSS symptoms, as well as global functioning (F's = 22.14–82.33, p b 0.001), with schizophrenia patients having more symptoms and lower functioning than both controls and relatives (p's b 0.001). Importantly, relatives and controls did not significantly differ for psychiatric and global functioning (p's = 0.09–0.32). Furthermore, relatives and controls did not differ on percentage of individuals with a lifetime Axis I disorder (X2(1) = 0.34, p = 0.56). 3.2. Diffusion tensor imaging measures 3.2.1. Classical whole-tract measures Four 2 hemisphere (left, right) by 3 group (schizophrenia, relatives, controls) repeated measures ANCOVAs on FA, MD, RD, and AD of the body of the fornix demonstrated no main effects of hemisphere (F's = 0.04–0.26, p's = 0.27–0.85), group (F's = 0.48–1.78, p's = 0.25–0.62, partial η2 = 0.01–0.04), or hemisphere by group interactions (F's = 0.40–1.94, p's = 0.15–0.67, partial η2 = 0.01–0.05). Four 2 hemisphere (left, right) by 3 group (schizophrenia, relatives, controls) repeated measures ANCOVAs on FA, MD, RD, and AD of the fimbria of the fornix demonstrated no main effects of hemisphere (F's = 0.02–1.34, p's = 0.25–0.89), group (F's = 0.495–1.48, p's = 0.24–0.61, partial η2 = 0.01–0.04), or hemisphere by group interactions (F's = 1.26–2.87, p's = 0.06–0.29, partial η2 = 0.01–0.08).
3.1. Participants Table 1 presents participant characteristics. Groups did not differ significantly for age (F(2, 73) = 0.05, p = 0.95), gender (X2(2) = 0.54, p = 0.77), or handedness (F(2) = 2.83, p = 0.24). Groups differed significantly for highest education level achieved (F(2, 73) = 4.55, p = 0.01), with relatives having a higher level of education compared to schizophrenia patients (p = 0.004), with no differences amongst other groups (p's = 0.13). However, groups did not differ for mothers' (F(2, 70) = 0.19, p = 0.82) or fathers' (F(2, 64) = 0.75, p = 0.48) education level. Groups differed on annual income (X2(6) = 34.54, p b 0.001), with schizophrenia patients having more individuals in lower income brackets than controls and relatives (X2's = 21.05–23.16, p b 0.001), but no differences between controls and relatives (X2(3) = 1.36, p = 0.72). Groups also differed on the vocabulary subtest of the WASI (F(2, 71) = 3.48, p = 0.04), with relatives having higher scores than schizophrenia patients, with no differences amongst other groups (p's = 0.10–0.297); however, groups did not differ on the matrix reasoning subtest (F(2, 71) = 0.52, p = 0.596).
3.2.2. Along-tract measures Twelve 2 group (schizophrenia vs. controls; relatives vs. controls, schizophrenia vs. relatives) repeated measures ANCOVAs on FA, MD, RD, and AD of the body of the fornix demonstrated no main effect of group or group by position along the tract interaction along the left body (F's = 0.06–6.83, p's = 0.098–1, partial η2 = b0.001–0.05) or right body (F's = 0.005–4.72, p's = 0.04–1, partial η2 b 0.001–0.05; Fig. 1) after FDR correction for multiple comparisons. For the right fimbria of the fornix, the repeated measures ANCOVAs demonstrated a significant group by position along the tract interaction, suggesting local increases in FA in relatives compared to controls (F(73) = 1.88, p b 0.001, partial η2 = 0.005) and schizophrenia patients (F(73) = 1.75, p = 0.002, partial η2 = 0.005). There were no significant main effects of group (F's = 0.030–3.48; p's = 0.42–0.92, partial η2 = b0.001–0.008). In both contrasts, an effect of the age covariate was found (F's = 4.22–4.33, p's = 0.001–0.002), but not the gender covariate (F = 0.09–1.01, p = 0.36–0.77). No other significant group or group by position along the tract interaction effects was
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
found for the FA, MD, RD, or AD for the left or right fimbria contrasts (F's = 0.003–5.59 p's = 0.27–1, partial η2 b 0.001–0.009). 3.3. Relationship between symptoms, functioning, and IQ with DTI metrics For comparison to previous literature, we assessed whether bilateral whole-tract FA values were associated with positive and negative symptoms from the PANSS or global functioning in schizophrenia patients and relatives and with cognition in all three groups. No significant associations were found between symptoms (r's = 0.20–0.36, p's = 0.10– 0.37; r's = − 0.06–0.16, p's = 0.24–0.77) or global functioning (r's = − 0.14 to − 0.22, p = 0.33–0.54; r's = 0.004–0.13, p's = 0.57–0.99) and FA values of the body or fimbria of the fornix in patients or relatives respectively. Similarly, no significant associations (correcting for multiple comparisons) were found between either the vocabulary or matrix reasoning subtests and FA values of the body or fimbria of the fornix in schizophrenia patients (r's = −0.17–0.08, p's = 0.47– 0.96), relatives (r's = − 0.49–0.20, p's = 0.02–0.93), or controls (r's = −0.12 to −0.33, p's = 0.10–0.57). 4. Discussion In this study we used sophisticated DTI analyses, including novel along-tract statistics along with whole-tract means to comprehensively assess two subdivisions of the fornix using multiple metrics that reflect WM microstructural organization of the fornix in schizophrenia patients, non-psychotic relatives, and controls. We did not find differences using classical whole-tract means; however, our advanced along-tract analysis suggests differences in relatives compared to controls and patients. Using along-tract statistics, local increased FA in the right fimbria was found in relatives compared to both controls and patients. Similar to two previous studies in relatives, we did not find differences in FA for whole-tract means in the fornix (Boos et al., 2013; Skudlarski et al., 2013). This finding in the absence of a global change can be interpreted both in the context of the methodological approach and in the context of neurobiology. First, a local difference could be obscured when averaging values along the length of the tract. Second, local differences could reflect focal changes in microstructural organization either in the fornix itself, or in surrounding structures such as the hippocampus and ventricular system. Such changes may reflect microarchitectural differences in WM, for example, in membrane density/cellularity and axonal orientation or macroscopic changes such as enlarged CSF spaces (Beaulieu, 2002; Jones et al., 2013). Increased FA is theoretically thought to represent greater WM coherence. Therefore, our finding of greater FA in the fornix, a key WM tract that connects to the hippocampus, could represent a compensatory mechanism to protect against psychosis in relatives. Alternatively, this finding could represent a maladaptive abnormality. Several studies in schizophrenia found greater FA in fibers connecting the temporal lobe (e.g. superior longitudinal fasciculus) in patients compared to controls (Alba-Ferrara and de Erausquin, 2013) and greater FA in these tracts were associated with hallucinations (Alba-Ferrara and de Erausquin, 2013). Gray matter studies have also found increases in temporal lobe regions in non-psychotic relatives compared to controls (Seidman et al., 2003; Goghari et al., 2007). Our classical whole-tract findings differ from previous studies of whole-tract measures that typically find differences between schizophrenia patients and controls (e.g., Kuroki et al., 2006; Fitzsimmons et al., 2014). One reason for the discrepancy between studies could be sample characteristics. Previous research has demonstrated that combinations of previous and current antipsychotic medications can obscure effects in schizophrenia patients (Goghari et al., 2013). Alternatively, methodological differences, may account for differences between studies.
439
Our study had a number of methodological strengths. We used DTI data collected at higher than average b-values to increase angular resolution, and 60 diffusion weighted gradient directions to improve signalto-noise ratio. Additionally, we used a population-based template to limit registration errors and, hence, CSF partial volume effects that confound similar analyses of the fornix were minimized. Moreover, we restricted our analysis to well-defined, uniform segments of the fornix, reducing intra-subject variability and improving the reliability of our along-tract analysis. Unlike in other studies, we accounted for variation in DTI metrics along the fornix using along-tract statistics, which also allowed us to visually confirm the basis of our findings. This increased the sensitivity of our analysis to both global and local differences if they were present. This study also has several limitations. One limitation of our study was we allowed multiple relatives from a family to participate to enhance recruitment; therefore, not maintaining the non-independence of every individual in the sample. Another limitation of our study was the small sample size. However, our sample size was similar to most other studies of schizophrenia patients and relatives (e.g., Camchong et al., 2009; Clark et al., 2011; Knochel et al., 2012). To compare to previous literature and for future meta-analyses, we calculated Cohen's d effect sizes for the classical whole-tract means. The effect sizes were negligible to moderate—moderate effect sizes are worthy of further investigation. Additionally, many of our group differences detected using along-tract statistics did not survive statistical correction, suggesting insufficient power to detect group differences. Larger sample sizes are recommended and will also allow for investigation of the relationship between brain and behavior necessary to interpret the direction of findings in patients and relatives. In summary, we found subtle effects in relatives of schizophrenia patients in the fornix compared to controls and patients. These findings underscore the utility of investigating along-tract measures to determine the nature of WM abnormalities in schizophrenia. Last, family members, who have some of the genes for the disorder, but do not manifest the disorder, may be an ideal sample to investigate successful compensatory neural and behavioral mechanisms that could be the focus of novel pharmacological and psychosocial interventions. Role of funding source This work was supported by a Canadian Institutes of Health Research Operating Grant and New Investigator Award to VMG, a European Union Seventh Framework Programme under grant agreement 279281 to TB, KU Leuven PFV/10/008 grant to LE and SS, and a KU Leuven International Mobility Bursary to LE. The funding sources had no further role in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. Contributors Vina Goghari and Louise Emsell designed the study and wrote the protocol. Vina Goghari, Louise Emsell, Thibo Billiet, and Stefan Sunaert managed the literature searches and analyses. Vina Goghari, Louise Emsell, and Thibo Billiet undertook the statistical analysis, and wrote the first draft of the manuscript. All authors contributed to and have approved the final manuscript. Conflict of interest No authors report any potential conflicts of interest. Acknowledgements We gratefully acknowledge Jennifer Prentice, Andrea Moir, Cameron Clark, and Irene Liu for their help with data collection. We would also like to thank Frederik Dekeyzer and the KU Leuven Biostatistics and Statistical Bioinformatics Centre for their statistical support.
References Abdul-Rahman, M.F., Qiu, A., Sim, K., 2011. Regionally specific white matter disruptions of fornix and cingulum in schizophrenia. PLoS One 6 (4), e18652. Alba-Ferrara, L.M., de Erausquin, G.A., 2013. What does anisotropy measure? Insights from increased and decreased anisotropy in selective fiber tracts in schizophrenia. Front. Integr. Neurosci. 7, 9. APA, 1994. Diagnostic and Statistical Manual of Mental Disorders: DSM-IV, 5th ed. American Psychiatric Publishing, Washington, DC.
440
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
Bach, M., Laun, F.B., Leemans, A., Tax, C.M., Biessels, G.J., Stieltjes, B., Maier-Hein, K.H., 2014. Methodological Considerations on Tract-based Spatial Statistics (TBSS). Neuroimage. Beaulieu, C., 2002. The basis of anisotropic water diffusion in the nervous system — a technical review. NMR Biomed. 15 (7–8), 435–455. Birchwood, M., Smith, J., Cochrane, R., Wetton, S., Copestake, S., 1990. The Social Functioning Scale. The development and validation of a new scale of social adjustment for use in family intervention programmes with schizophrenic patients. Br. J. Psychiatry 157, 853–859. Boos, H.B.M., Mandl, R.C.W., van Haren, N.E.M., Cahn, W., van Baal, G.C.M., Kahn, R.S., Pol, H.E.H., 2013. Tract-based diffusion tensor imaging in patients with schizophrenia and their non-psychotic siblings. Eur. Neuropsychopharmacol. 23 (4), 295–304. Camchong, J., Lim, K.O., Sponheim, S.R., Macdonald, A.W., 2009. Frontal white matter integrity as an endophenotype for schizophrenia: diffusion tensor imaging in monozygotic twins and patients' nonpsychotic relatives. Front. Hum. Neurosci. 3, 35. Clark, K.A., Nuechterlein, K.H., Asarnow, R.F., Hamilton, L.S., Phillips, O.R., Hageman, N.S., Woods, R.P., Alger, J.R., Toga, A.W., Narr, K.L., 2011. Mean diffusivity and fractional anisotropy as indicators of disease and genetic liability to schizophrenia. J. Psychiatr. Res. 45 (7), 980–988. Colby, J.B., Soderberg, L., Lebel, C., Dinov, I.D., Thompson, P.M., Sowell, E.R., 2012. Along-tract statistics allow for enhanced tractography analysis. Neuroimage 59 (4), 3227–3242. Emsell, L., Leemans, A., Langan, C., Van Hecke, W., Barker, G.J., McCarthy, P., Jeurissen, B., Sijbers, J., Sunaert, S., Cannon, D.M., McDonald, C., 2013. Limbic and callosal white matter changes in euthymic bipolar I disorder: an advanced diffusion magnetic resonance imaging tractography study. Biol. Psychiatry 73 (2), 194–201. Fitzsimmons, J., Hamoda, H.M., Swisher, T., Terry, D., Rosenberger, G., Seidman, L.J., Goldstein, J., Mesholam-Gately, R., Petryshen, T., Wojcik, J., Kikinis, R., Kubicki, M., 2014. Diffusion tensor imaging study of the fornix in first episode schizophrenia and in healthy controls. Schizophr. Res. 156 (2–3), 157–160. Goghari, V.M., Rehm, K., Carter, C.S., MacDonald 3rd, A.W., 2007. Regionally specific cortical thinning and gray matter abnormalities in the healthy relatives of schizophrenia patients. Cereb. Cortex 17 (2), 415–424. Goghari, V.M., Smith, G.N., Honer, W.G., Kopala, L.C., Thornton, A.E., Su, W., Macewan, G.W., Lang, D.J., 2013. Effects of eight weeks of atypical antipsychotic treatment on middle frontal thickness in drug-naive first-episode psychosis patients. Schizophr. Res. 149 (1–3), 149–155. Guo, W., Liu, F., Liu, Z., Gao, K., Xiao, C., Chen, H., Zhao, J., 2012. Right lateralized white matter abnormalities in first-episode, drug-naive paranoid schizophrenia. Neurosci. Lett. 531 (1), 5–9. Jbabdi, S., Johansen-Berg, H., 2011. Tractography: where do we go from here? Brain Connect. 1 (3), 169–183.
Jones, D.K., Knosche, T.R., Turner, R., 2013. White matter integrity, fiber count, and other fallacies: the do's and don'ts of diffusion MRI. Neuroimage 73, 239–254. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 13 (2), 261–276. Kendler, K.S., Lieberman, J.A., Walsh, D., 1989. The Structured Interview for Schizotypy (SIS): a preliminary report. Schizophr. Bull. 15 (4), 559–571. Knochel, C., O'Dwyer, L., Alves, G., Reinke, B., Magerkurth, J., Rotarska-Jagiela, A., Prvulovic, D., Hampel, H., Linden, D.E., Oertel-Knochel, V., 2012. Association between white matter fiber integrity and subclinical psychotic symptoms in schizophrenia patients and unaffected relatives. Schizophr. Res. 140 (1–3), 129–135. Kuroki, N., Kubicki, M., Nestor, P.G., Salisbury, D.F., Park, H.J., Levitt, J.J., Woolston, S., Frumin, M., Niznikiewicz, M., Westin, C.F., Maier, S.E., McCarley, R.W., Shenton, M.E., 2006. Fornix integrity and hippocampal volume in male schizophrenic patients. Biol. Psychiatry 60 (1), 22–31. Lebel, C., Gee, M., Camicioli, R., Wieler, M., Martin, W., Beaulieu, C., 2012. Diffusion tensor imaging of white matter tract evolution over the lifespan. Neuroimage 60 (1), 340–352. Leemans, A., Jeurissen, B., Sijbers, J., Jones, D.K., 2009. ExploreDTI: a graphical toolbox for processing, analyzing, and visualizing diffusion MR data. 17th Annual Meeting of Intl Soc Mag Reson Med, p. 3537. Sacher, J., Neumann, J., Okon-Singer, H., Gotowiec, S., Villringer, A., 2013. Sexual dimorphism in the human brain: evidence from neuroimaging. Magn. Reson. Imaging 31 (3), 366–375. Seidman, L.J., Pantelis, C., Keshavan, M.S., Faraone, S.V., Goldstein, J.M., Horton, N.J., Makris, N., Falkai, P., Caviness, V.S., Tsuang, M.T., 2003. A review and new report of medial temporal lobe dysfunction as a vulnerability indicator for schizophrenia: a magnetic resonance imaging morphometric family study of the parahippocampal gyrus. Schizophr. Bull. 29 (4), 803–830. Skudlarski, P., Schretlen, D.J., Thaker, G.K., Stevens, M.C., Keshavan, M.S., Sweeney, J.A., Tamminga, C.A., Clementz, B.A., O'Neil, K., Pearlson, G.D., 2013. Diffusion tensor imaging white matter endophenotypes in patients with schizophrenia or psychotic bipolar disorder and their relatives. Am. J. Psychiatry 170 (8), 886–898. Van Hecke, W., Leemans, A., D'Agostino, E., De Backer, S., Vandervliet, E., Parizel, P.M., Sijbers, J., 2007. Nonrigid coregistration of diffusion tensor images using a viscous fluid model and mutual information. IEEE Trans. Med. Imaging 26 (11), 1598–1612. Van Hecke, W., Sijbers, J., D'Agostino, E., Maes, F., De Backer, S., Vandervliet, E., Parizel, P.M., Leemans, A., 2008. On the construction of an inter-subject diffusion tensor magnetic resonance atlas of the healthy human brain. Neuroimage 43 (1), 69–80. Wechsler, D., 1999. Wechsler Abbreviated Scale of Intelligence. Pearson Education, Inc., San Antonio, TX.
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
A diffusion tensor imaging family study of the fornix in schizophrenia Vina M. Goghari a,⁎, Thibo Billiet b, Stefan Sunaert b, Louise Emsell b a b
Clinical Neuroscience of Schizophrenia (CNS) Laboratory, Department of Psychology, Hotchkiss Brain Institute, University of Calgary, 2500 University Drive, NW Calgary, AB T2N 1N4, Canada Translational MRI, Department of Imaging and Pathology, KU Leuven and Radiology, University Hospitals Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 2 August 2014 Received in revised form 18 September 2014 Accepted 22 September 2014 Available online 12 October 2014 Keywords: Psychosis Tractography Fornix Family study Hippocampus White matter
a b s t r a c t Diffusion tensor imaging (DTI) studies suggest abnormalities in the white matter microstructure of the fornix in schizophrenia patients. Research evaluating schizophrenia patient and relatives also suggests that the white matter microstructure of the fornix is heritable. However, previous studies have been hindered by limited DTI methodology. Therefore, the goal of this study was to assess whether fornix abnormalities were related to the genetic liability for schizophrenia using the novel methodological approach of assessing multiple metrics of along-tract measurements, in addition to whole-tract means. Twenty-five schizophrenia patients, 24 adult non-psychotic first-degree biological relatives, and 27 community controls underwent neuroimaging. No group differences were found for any of the DTI metrics using the classical whole-tract measures of the fornix. Along-tract analysis detected local increases in fractional anisotropy (FA) in the right fimbria of the fornix for relatives compared to patients and controls corrected for false discovery rate. No significant associations were found between symptoms, global functioning, or IQ and whole-tract FA means in schizophrenia patients or relatives. Increased FA in nonpsychotic relatives could represent a compensatory mechanism to guard against psychosis or an abnormality associated with the genetic liability for the disorder. These findings underscore the importance of obtaining alongtract measurements, in addition to whole-tract measurements to fully understand white matter abnormalities in schizophrenia. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Diffusion tensor imaging (DTI) metrics differentiate between schizophrenia patients and controls, and furthermore, appear to be useful vulnerability markers for schizophrenia (e.g., Camchong et al., 2009; Clark et al., 2011; Knochel et al., 2012; Skudlarski et al., 2013). Additionally, these measures are associated with symptoms of the disorder in both patients and relatives (Knochel et al., 2012). Although most white matter (WM) tracts have been investigated, most DTI studies of relatives have not assessed the fornix. The fornix is a tract of key interest in schizophrenia, given it connects the hippocampus to the hypothalamus. Several studies in schizophrenia patients have found WM abnormalities in the fornix using voxel-based (e.g., Guo et al., 2012) and tract-based DTI analyses (e.g., Fitzsimmons et al., 2014). While tractography itself is not a quantitative technique (it is a way of virtually reconstructing WM fiber-like pathways), it is often used synonymously with the method used to extract mean quantitative values of tracts generated from tractography (Jbabdi and Johansen-Berg, 2011). However, it is known that DTI metrics vary along the length of tracts; therefore, taking a simple mean of all these ⁎ Corresponding author at: Department of Psychology, University of Calgary, 2500 University Drive, NW Calgary, AB T2N 1N4, Canada. Tel.: +1 403 210 7344; fax: +1 403 282 8249. E-mail address: [email protected] (V.M. Goghari).
http://dx.doi.org/10.1016/j.schres.2014.09.037 0920-9964/© 2014 Elsevier B.V. All rights reserved.
values, as in classical tractography analysis, may be insufficient for capturing more subtle local differences (Colby et al., 2012). A more sophisticated approach that accounts for this variation is along-tract analysis, whereby focal differences in parameters manifest as local offsets between along-tract profiles (i.e., the plot of metric value at predetermined points along a tract) and whole-tract differences manifest as a global off-set between tract profiles. Abdul-Rahman et al. (2011) used along tract-based analysis and found specific loci for fractional anisotropy (FA) reductions and axial diffusivity (AD) and radial diffusivity (RD) increases in the left fornix. Furthermore, this study found decreased FA in the specific part of the left fornix that was related to increased psychotic symptoms in patients (Abdul-Rahman et al., 2011). Along-tract methods have not been used in any family studies of schizophrenia patients. Using traditional methods, two large scale studies, including over 100 relatives and controls each, did not find FA differences in the fornix in relatives versus controls (Boos et al., 2013; Skudlarski et al., 2013). However, one study found that FA values of the fornix were heritable in schizophrenia patients and relatives (Skudlarski et al., 2013), suggesting that further investigation of the fornix with more sophisticated DTI methods, in relatives, is warranted. Most previous studies investigating WM changes in schizophrenia have used voxel-based analysis, typically of a highly restricted WM skeleton (e.g., tract based spatial statistics), which is not an optimal approach for investigating a narrow tract such as the fornix (Bach et al.,
436
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
2014). Other studies have used whole-tract mean values from tractography; thereby, losing anatomical specificity. The goal of this investigation therefore, was to use advanced DTI analysis to examine along-tract microstructural differences, in addition to whole-tract mean values, including measures of FA, mean diffusivity (MD), RD, and AD, in two subdivisions of the fornix in a family study of schizophrenia. We expected significant differences between all groups, with greater along-tract differences than whole-tract mean differences. We hypothesized that the largest differences would be found between patients and controls with non-psychotic relatives being intermediate. In terms of the individual metrics, we expected reduced FA and increased MD, RD, and AD in patients and relatives compared to controls in both parts of the fornix. 2. Materials and methods
families in which two relatives participated and one family in which three relatives participated. Community controls were recruited through flyers and advertisements around the community. The University of Calgary Ethics Board approved the protocol and informed written consent was obtained. Inclusion criteria for all participants included: (1) age 18–65; (2) minimum IQ of 70; (3) no current diagnosis of drug/alcohol dependence/abuse; (4) no history of head injury or being unconscious for more than 20 min; (5) no history of electroconvulsive therapy; and (6) no history of stroke or other neurological condition. Further criteria for inclusion of first-degree relatives were no lifetime diagnosis of a psychotic disorder or bipolar disorder, or history of anti-psychotic medication use. Further criteria for inclusion of community controls were no personal or family history of a psychotic disorder or bipolar disorder, or personal use of an anti-psychotic medication.
2.1. Participants 2.2. Diagnosis and assessment A total of 76 individuals participated: 25 schizophrenia or schizoaffective patients (hereafter referred to as schizophrenia patients), 24 adult non-psychotic first-degree biological relatives, and 27 community controls. Demographic characteristics are shown in Table 1. Schizophrenia patients were recruited through outpatient clinics and through community support programs in Calgary, Canada. Research staff identified first-degree biological relatives by completing a pedigree with the proband. Not all probands met recruitment criteria for the study; however, to enhance the sample, all first-degree biological relatives of schizophrenia patients that met recruitment criteria were included. Seven relatives were related to probands in this sample and there were two
Participants were interviewed using the Structured Clinical Interview for DSM-IV Axis I Disorders. The Structured Interview for Schizotypy, with supplemental questions, was used to measure Axis II Cluster A disorders in relatives and controls (Kendler et al., 1989). Diagnoses were assigned according to DSM-IV-TR criteria via case conferences. No relatives or controls met criteria for a Cluster A disorder. Symptoms and functioning were measured using the Positive and Negative Syndrome Scale (PANSS; Kay et al., 1987), Social Functioning Scale (Birchwood et al., 1990), the Global Assessment of Functioning (GAF) Scale (APA, 1994). Lastly, the vocabulary and matrix reasoning subtests of the Wechsler
Table 1 Participant characteristics: demographics, IQ measures, symptoms, functioning, and medication usage.
N Age Gender (% female) Born in Canada (%) Education (years completed) Annual income (%)1 $0–$30,000 $30,000–$50,000 $50,000–$95,000 $95,000+ Maternal education (years completed) Paternal education (years completed) Matrix reasoning raw score Vocabulary raw score Handedness (% right handed) PANSS negative: range PANSS positive: range PANSS general: range Global assessment of functioning: range Social functioning scale: range Axis I (% with any lifetime diagnosis) Relative status — parent:sibling:offspring Anti-psychotic (atypical, typical, both; % on) Anti-depressants (% on) Mood stabilizer (% on) Anti-anxiety (% on) Anti-parkinson (% on) Other psychiatric (% on)
Schizophrenia
Relative
Control
25 41.3 (10.8) 48 92 14.2 (3.0)a
24 40.2 (15.0) 58 88 16.4 (2.6)
27 40.7 (11.1) 52 81 15.3 (2.4)
64 20 12 4 13.7 (2.8) 14.0 (3.0) 26.3 (2.8) 57.1 (5.9)a 92 11.9 (3.6): 7–19 14.4 (5.3): 7–24 26.8 (6.5): 16–39 53.4 (14.9): 30–83a,c
4 17 54 25 13.1 (3.8) 12.6 (4.0) 27.6 (3.1) 62.2 (5.8) 79 7.8 (1.1): 7–11b 8.8 (1.9): 7–14b 20.5 (3.6): 16–29b 81.8 (5.7): 63–88
– N/A 96, 4, 4 40 12 8 4 8
332 9:12:3 0, 0, 0 8 0 4 0 4
8 19 39 35 13.5 (3.4) 13.7 (4.7) 27.0 (5.6) 59.1 (7.8) 96 7.2 (0.6): 7–10b 7.9 (1.3): 7–11b 18.1 (37): 16–33b 84.7 (5.1): 73–95 793.5 (53.1): 701–883 263 N/A 0, 0, 0 8 0 0 0 0
Note. Mean and standard deviation presented where appropriate. The following notations were used for non-medication variables: a Less than relatives. b Less than patients. c Less than controls. 1 Overall chi-square demonstrated patients had more individuals in lower income brackets than controls or relatives. 2 Some relatives had more than one lifetime Axis I disorder: 1 with alcohol dependence, 4 with alcohol abuse, 1 with cannabis abuse, 5 with major depressive disorder, 1 with dysthymia, 2 with post-traumatic stress disorder, 1 with panic disorder without agoraphobia, and 1 with generalized anxiety disorder. 3 Some controls had more than one lifetime Axis I disorder: 1 with alcohol dependence, 1 with cocaine dependence, 3 with alcohol abuse, 1 with hallucinogen abuse, 1 with cannabis dependence, 3 with major depressive disorder, and 1 with social anxiety disorder.
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
(a)
Body
437
(b)
Fimbria
(i)
Fornix
FA
MD
(ii)
Le body
RD (iii)
Le fimbria
AD
Le
Right Controls
Le Relaves
Right
Schizophrenia
Fig. 1. Diffusion tensor imaging of the fornix in schizophrenia patients, non-psychotic relatives, and controls. (a) Changes in DTI parameters along the length of the fornix body and fimbria subdivisions between schizophrenia patients, non-psychotic relatives, and community controls. The x-axis describes the position along the tract, beginning to end at intervals of 1 mm. In the body divisions, it represents points 0 to 24, and in the left/right fimbria points 0 to 74/78. The along-tract analysis assessed both changes in the mean offset (i.e., tract mean) between the profiles (i.e. differences between the red, green and blue lines), and local (i.e., tract position) differences. The significant local difference in FA between relatives and patients and controls can be seen in the mid-fimbria. (b) The three-dimensional tract masks used to represent the whole fornix (i), colored according to the principal direction of diffusion, and to generate its subdivisions, the body (left hemisphere) (ii) and fimbria (left hemisphere) (iii), both overlaid on the population atlas color FA map. Abbreviations: FA: fractional anisotropy, MD: mean diffusivity, AD: axial diffusivity, and RD: radial diffusivity. FA is a fraction, without units. Units of MD, RD & AD = mm2/s.
Abbreviated Scale of Intelligence were used to estimate IQ (WASI; Wechsler, 1999).
2.3. Diffusion tensor imaging High-angular resolution diffusion (HARDI) MRI data were acquired (3 T GE) along 60 gradient directions with b-value 1300 s/mm2 and corrected for motion and distortion using ExploreDTI (Leemans et al., 2009). DTI parameters FA, MD, RD, and AD were estimated. A population-based template was constructed from the participant datasets, to which all diffusion tensor maps were registered (Van Hecke et al., 2008). This approach utilized non-affine registration and tensor reorientation, in order to correctly preserve the orientation information captured in the diffusion signal (Van Hecke et al., 2007). Parameter maps were obtained in atlas space. The fornix was isolated in template space using targeted deterministic tractography in ExploreDTI (Emsell et al., 2013). Seed-based tracking was initiated in the fornix body, proceeded with a step size of 1 mm and terminated in voxels with an FA of N 0.12. Spurious fibers not belonging to the fornix were removed by tract selection regions of interest. Subdivision of the whole tract into left/right body and left/right fimbria was performed to ensure consistent streamline lengths for along-tract sampling. The body and fimbria subdivisions were defined as the point at which the whole body diverged into two tracts (i.e., the fimbria). The anterior boundary of the body was defined as the point at which it divided into the two
fornix columns (Fig. 1b). In-house software was used to calculate parameter values at 1 mm intervals along the length of the tract masks for each subject in template space. For the left and right body, 24 points were assessed in each tract. For the left fimbria, 74 points were assessed, and for the right fimbria, 78 points. For the along-tract analysis, average parameter values were taken at each of these evenly spaced points along the reconstructed tractography streamlines comprising the four fornix segments (i.e., left/right body and left/right fimbria) in each participant separately. Classical whole-tract means were also calculated by dividing the total metric value at every point by the number of points per tract in each participant.
2.4. Statistical analyses Statistical analysis softwares used were SPSS v 21 and MATLAB version 2011b. Age and gender were used as covariates in all analyses, as there is evidence from other studies that DTI parameters may vary as a function of age (Lebel et al., 2012) and gender (Sacher et al., 2013). Greenhouse–Geisser correction is reported for analyses of covariance (ANCOVAs). Partial eta-squared effect sizes (η2) are also provided. For the classical whole-tract means, four 2 hemisphere (left, right) by 3 group ANCOVAs were used to assess for the effect of group on FA, MD, RD, and AD separately for the body and fimbria of the fornix. For comparison to previous literature and inclusion in meta-analyses, Cohen's d effect sizes are presented for all group comparisons of the whole-
438
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
Table 2 Raw classical whole-tract mean values and absolute Cohen's d effect size values for group comparisons for DTI metrics of body and fimbria of the fornix.
Body FA left MD left RD left AD left FA right MD right RD right AD right Fimbria FA left MD left RD left AD left FA right MD right RD right AD right
Schizophrenia
Relative
Control
Schizophrenia vs control Cohen's d
Relative vs control Cohen's d
Schizophrenia vs relative Cohen's d
0.43 (0.06) 0.0018 (0.0002) 0.0015 (0.0002) 0.0023 (0.0002) 0.43 (0.06) 0.0018 (0.0002) 0.0015 (0.0002) 0.0022 (0.0001)
0.45 (0.03) 0.0017 (0.0001) 0.0014 (0.0001) 0.0022 (0.0001) 0.43 (0.05) 0.0017 (0.0001) 0.0015 (0.0001) 0.0022 (0.0001)
0.43 (0.04) 0.0017 (0.0001) 0.0015 (0.0001) 0.0022 (0.0001) 0.43 (0.04) 0.0017 (0.0001) 0.0015 (0.0001) 0.0022 (0.0001)
0.05 0.16 0.13 0.20 0.01 0.18 0.15 0.25
0.53 0.40 0.44 0.31 0.10 0.22 0.20 0.24
0.33 0.52 0.52 0.50 0.09 0.38 0.33 0.45
0.39 (0.03) 0.0014 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001) 0.41 (0.03) 0.0013 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001)
0.40 (0.04) 0.0013 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001) 0.41 (0.04) 0.0013 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001)
0.38 (0.04) 0.0014 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001) 0.40 (0.03) 0.0013 (0.0001) 0.0012 (0.0001) 0.0017 (0.0001)
0.27 0.01 0.04 0.11 0.07 0.13 0.10 0.19
0.48 0.40 0.04 0.32 0.30 0.16 0.18 0.11
0.25 0.43 0.43 0.43 0.23 0.28 0.27 0.29
Mean and standard deviation. Cohen's d was calculated using as many decimal places as available. FA = fractional anisotropy, MD = mean diffusivity, RD = radial diffusivity, AD = axial diffusivity. Units of MD, RD and AD = mm2/s.
tract raw values in Table 2. Follow-up testing was only conducted if the 3 group ANCOVA was significant to control for multiple comparisons. Along-tract analysis was used to detect both local differences in DTI measures and global offset of the tract profiles. Whereas the classical approach assesses the difference between 3 points (i.e. the tract mean of group 1, tract mean group 2, tract mean group 3), in the along-tract approach, there is a repeated measures term of ‘position’ as a result of variance in metrics at each point along the tract. The along-tract approach accounts for this variance by investigating the offset between 3 spatially varying curves (Colby et al., 2012). This requires a different statistical model with many more terms (i.e., tract points). Twelve ANCOVAs (4 metrics × 3 two-group contrasts) with group, position along the tract, and group by position interaction were used to assess for the effect of group on FA, MD, RD, and AD separately for the body and fimbria of the fornix. As this was a novel method, we a priori directly compared each two groups (patients vs. controls, relatives vs. controls, relatives vs. patients). False discovery rate (FDR) correction was applied to all along-tract analyses as a whole to control for multiple comparisons. Lastly, partial correlations were conducted to assess the relationship between FA and symptoms, global functioning, and IQ.
3. Results
Groups differed in their positive, negative, and general PANSS symptoms, as well as global functioning (F's = 22.14–82.33, p b 0.001), with schizophrenia patients having more symptoms and lower functioning than both controls and relatives (p's b 0.001). Importantly, relatives and controls did not significantly differ for psychiatric and global functioning (p's = 0.09–0.32). Furthermore, relatives and controls did not differ on percentage of individuals with a lifetime Axis I disorder (X2(1) = 0.34, p = 0.56). 3.2. Diffusion tensor imaging measures 3.2.1. Classical whole-tract measures Four 2 hemisphere (left, right) by 3 group (schizophrenia, relatives, controls) repeated measures ANCOVAs on FA, MD, RD, and AD of the body of the fornix demonstrated no main effects of hemisphere (F's = 0.04–0.26, p's = 0.27–0.85), group (F's = 0.48–1.78, p's = 0.25–0.62, partial η2 = 0.01–0.04), or hemisphere by group interactions (F's = 0.40–1.94, p's = 0.15–0.67, partial η2 = 0.01–0.05). Four 2 hemisphere (left, right) by 3 group (schizophrenia, relatives, controls) repeated measures ANCOVAs on FA, MD, RD, and AD of the fimbria of the fornix demonstrated no main effects of hemisphere (F's = 0.02–1.34, p's = 0.25–0.89), group (F's = 0.495–1.48, p's = 0.24–0.61, partial η2 = 0.01–0.04), or hemisphere by group interactions (F's = 1.26–2.87, p's = 0.06–0.29, partial η2 = 0.01–0.08).
3.1. Participants Table 1 presents participant characteristics. Groups did not differ significantly for age (F(2, 73) = 0.05, p = 0.95), gender (X2(2) = 0.54, p = 0.77), or handedness (F(2) = 2.83, p = 0.24). Groups differed significantly for highest education level achieved (F(2, 73) = 4.55, p = 0.01), with relatives having a higher level of education compared to schizophrenia patients (p = 0.004), with no differences amongst other groups (p's = 0.13). However, groups did not differ for mothers' (F(2, 70) = 0.19, p = 0.82) or fathers' (F(2, 64) = 0.75, p = 0.48) education level. Groups differed on annual income (X2(6) = 34.54, p b 0.001), with schizophrenia patients having more individuals in lower income brackets than controls and relatives (X2's = 21.05–23.16, p b 0.001), but no differences between controls and relatives (X2(3) = 1.36, p = 0.72). Groups also differed on the vocabulary subtest of the WASI (F(2, 71) = 3.48, p = 0.04), with relatives having higher scores than schizophrenia patients, with no differences amongst other groups (p's = 0.10–0.297); however, groups did not differ on the matrix reasoning subtest (F(2, 71) = 0.52, p = 0.596).
3.2.2. Along-tract measures Twelve 2 group (schizophrenia vs. controls; relatives vs. controls, schizophrenia vs. relatives) repeated measures ANCOVAs on FA, MD, RD, and AD of the body of the fornix demonstrated no main effect of group or group by position along the tract interaction along the left body (F's = 0.06–6.83, p's = 0.098–1, partial η2 = b0.001–0.05) or right body (F's = 0.005–4.72, p's = 0.04–1, partial η2 b 0.001–0.05; Fig. 1) after FDR correction for multiple comparisons. For the right fimbria of the fornix, the repeated measures ANCOVAs demonstrated a significant group by position along the tract interaction, suggesting local increases in FA in relatives compared to controls (F(73) = 1.88, p b 0.001, partial η2 = 0.005) and schizophrenia patients (F(73) = 1.75, p = 0.002, partial η2 = 0.005). There were no significant main effects of group (F's = 0.030–3.48; p's = 0.42–0.92, partial η2 = b0.001–0.008). In both contrasts, an effect of the age covariate was found (F's = 4.22–4.33, p's = 0.001–0.002), but not the gender covariate (F = 0.09–1.01, p = 0.36–0.77). No other significant group or group by position along the tract interaction effects was
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
found for the FA, MD, RD, or AD for the left or right fimbria contrasts (F's = 0.003–5.59 p's = 0.27–1, partial η2 b 0.001–0.009). 3.3. Relationship between symptoms, functioning, and IQ with DTI metrics For comparison to previous literature, we assessed whether bilateral whole-tract FA values were associated with positive and negative symptoms from the PANSS or global functioning in schizophrenia patients and relatives and with cognition in all three groups. No significant associations were found between symptoms (r's = 0.20–0.36, p's = 0.10– 0.37; r's = − 0.06–0.16, p's = 0.24–0.77) or global functioning (r's = − 0.14 to − 0.22, p = 0.33–0.54; r's = 0.004–0.13, p's = 0.57–0.99) and FA values of the body or fimbria of the fornix in patients or relatives respectively. Similarly, no significant associations (correcting for multiple comparisons) were found between either the vocabulary or matrix reasoning subtests and FA values of the body or fimbria of the fornix in schizophrenia patients (r's = −0.17–0.08, p's = 0.47– 0.96), relatives (r's = − 0.49–0.20, p's = 0.02–0.93), or controls (r's = −0.12 to −0.33, p's = 0.10–0.57). 4. Discussion In this study we used sophisticated DTI analyses, including novel along-tract statistics along with whole-tract means to comprehensively assess two subdivisions of the fornix using multiple metrics that reflect WM microstructural organization of the fornix in schizophrenia patients, non-psychotic relatives, and controls. We did not find differences using classical whole-tract means; however, our advanced along-tract analysis suggests differences in relatives compared to controls and patients. Using along-tract statistics, local increased FA in the right fimbria was found in relatives compared to both controls and patients. Similar to two previous studies in relatives, we did not find differences in FA for whole-tract means in the fornix (Boos et al., 2013; Skudlarski et al., 2013). This finding in the absence of a global change can be interpreted both in the context of the methodological approach and in the context of neurobiology. First, a local difference could be obscured when averaging values along the length of the tract. Second, local differences could reflect focal changes in microstructural organization either in the fornix itself, or in surrounding structures such as the hippocampus and ventricular system. Such changes may reflect microarchitectural differences in WM, for example, in membrane density/cellularity and axonal orientation or macroscopic changes such as enlarged CSF spaces (Beaulieu, 2002; Jones et al., 2013). Increased FA is theoretically thought to represent greater WM coherence. Therefore, our finding of greater FA in the fornix, a key WM tract that connects to the hippocampus, could represent a compensatory mechanism to protect against psychosis in relatives. Alternatively, this finding could represent a maladaptive abnormality. Several studies in schizophrenia found greater FA in fibers connecting the temporal lobe (e.g. superior longitudinal fasciculus) in patients compared to controls (Alba-Ferrara and de Erausquin, 2013) and greater FA in these tracts were associated with hallucinations (Alba-Ferrara and de Erausquin, 2013). Gray matter studies have also found increases in temporal lobe regions in non-psychotic relatives compared to controls (Seidman et al., 2003; Goghari et al., 2007). Our classical whole-tract findings differ from previous studies of whole-tract measures that typically find differences between schizophrenia patients and controls (e.g., Kuroki et al., 2006; Fitzsimmons et al., 2014). One reason for the discrepancy between studies could be sample characteristics. Previous research has demonstrated that combinations of previous and current antipsychotic medications can obscure effects in schizophrenia patients (Goghari et al., 2013). Alternatively, methodological differences, may account for differences between studies.
439
Our study had a number of methodological strengths. We used DTI data collected at higher than average b-values to increase angular resolution, and 60 diffusion weighted gradient directions to improve signalto-noise ratio. Additionally, we used a population-based template to limit registration errors and, hence, CSF partial volume effects that confound similar analyses of the fornix were minimized. Moreover, we restricted our analysis to well-defined, uniform segments of the fornix, reducing intra-subject variability and improving the reliability of our along-tract analysis. Unlike in other studies, we accounted for variation in DTI metrics along the fornix using along-tract statistics, which also allowed us to visually confirm the basis of our findings. This increased the sensitivity of our analysis to both global and local differences if they were present. This study also has several limitations. One limitation of our study was we allowed multiple relatives from a family to participate to enhance recruitment; therefore, not maintaining the non-independence of every individual in the sample. Another limitation of our study was the small sample size. However, our sample size was similar to most other studies of schizophrenia patients and relatives (e.g., Camchong et al., 2009; Clark et al., 2011; Knochel et al., 2012). To compare to previous literature and for future meta-analyses, we calculated Cohen's d effect sizes for the classical whole-tract means. The effect sizes were negligible to moderate—moderate effect sizes are worthy of further investigation. Additionally, many of our group differences detected using along-tract statistics did not survive statistical correction, suggesting insufficient power to detect group differences. Larger sample sizes are recommended and will also allow for investigation of the relationship between brain and behavior necessary to interpret the direction of findings in patients and relatives. In summary, we found subtle effects in relatives of schizophrenia patients in the fornix compared to controls and patients. These findings underscore the utility of investigating along-tract measures to determine the nature of WM abnormalities in schizophrenia. Last, family members, who have some of the genes for the disorder, but do not manifest the disorder, may be an ideal sample to investigate successful compensatory neural and behavioral mechanisms that could be the focus of novel pharmacological and psychosocial interventions. Role of funding source This work was supported by a Canadian Institutes of Health Research Operating Grant and New Investigator Award to VMG, a European Union Seventh Framework Programme under grant agreement 279281 to TB, KU Leuven PFV/10/008 grant to LE and SS, and a KU Leuven International Mobility Bursary to LE. The funding sources had no further role in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. Contributors Vina Goghari and Louise Emsell designed the study and wrote the protocol. Vina Goghari, Louise Emsell, Thibo Billiet, and Stefan Sunaert managed the literature searches and analyses. Vina Goghari, Louise Emsell, and Thibo Billiet undertook the statistical analysis, and wrote the first draft of the manuscript. All authors contributed to and have approved the final manuscript. Conflict of interest No authors report any potential conflicts of interest. Acknowledgements We gratefully acknowledge Jennifer Prentice, Andrea Moir, Cameron Clark, and Irene Liu for their help with data collection. We would also like to thank Frederik Dekeyzer and the KU Leuven Biostatistics and Statistical Bioinformatics Centre for their statistical support.
References Abdul-Rahman, M.F., Qiu, A., Sim, K., 2011. Regionally specific white matter disruptions of fornix and cingulum in schizophrenia. PLoS One 6 (4), e18652. Alba-Ferrara, L.M., de Erausquin, G.A., 2013. What does anisotropy measure? Insights from increased and decreased anisotropy in selective fiber tracts in schizophrenia. Front. Integr. Neurosci. 7, 9. APA, 1994. Diagnostic and Statistical Manual of Mental Disorders: DSM-IV, 5th ed. American Psychiatric Publishing, Washington, DC.
440
V.M. Goghari et al. / Schizophrenia Research 159 (2014) 435–440
Bach, M., Laun, F.B., Leemans, A., Tax, C.M., Biessels, G.J., Stieltjes, B., Maier-Hein, K.H., 2014. Methodological Considerations on Tract-based Spatial Statistics (TBSS). Neuroimage. Beaulieu, C., 2002. The basis of anisotropic water diffusion in the nervous system — a technical review. NMR Biomed. 15 (7–8), 435–455. Birchwood, M., Smith, J., Cochrane, R., Wetton, S., Copestake, S., 1990. The Social Functioning Scale. The development and validation of a new scale of social adjustment for use in family intervention programmes with schizophrenic patients. Br. J. Psychiatry 157, 853–859. Boos, H.B.M., Mandl, R.C.W., van Haren, N.E.M., Cahn, W., van Baal, G.C.M., Kahn, R.S., Pol, H.E.H., 2013. Tract-based diffusion tensor imaging in patients with schizophrenia and their non-psychotic siblings. Eur. Neuropsychopharmacol. 23 (4), 295–304. Camchong, J., Lim, K.O., Sponheim, S.R., Macdonald, A.W., 2009. Frontal white matter integrity as an endophenotype for schizophrenia: diffusion tensor imaging in monozygotic twins and patients' nonpsychotic relatives. Front. Hum. Neurosci. 3, 35. Clark, K.A., Nuechterlein, K.H., Asarnow, R.F., Hamilton, L.S., Phillips, O.R., Hageman, N.S., Woods, R.P., Alger, J.R., Toga, A.W., Narr, K.L., 2011. Mean diffusivity and fractional anisotropy as indicators of disease and genetic liability to schizophrenia. J. Psychiatr. Res. 45 (7), 980–988. Colby, J.B., Soderberg, L., Lebel, C., Dinov, I.D., Thompson, P.M., Sowell, E.R., 2012. Along-tract statistics allow for enhanced tractography analysis. Neuroimage 59 (4), 3227–3242. Emsell, L., Leemans, A., Langan, C., Van Hecke, W., Barker, G.J., McCarthy, P., Jeurissen, B., Sijbers, J., Sunaert, S., Cannon, D.M., McDonald, C., 2013. Limbic and callosal white matter changes in euthymic bipolar I disorder: an advanced diffusion magnetic resonance imaging tractography study. Biol. Psychiatry 73 (2), 194–201. Fitzsimmons, J., Hamoda, H.M., Swisher, T., Terry, D., Rosenberger, G., Seidman, L.J., Goldstein, J., Mesholam-Gately, R., Petryshen, T., Wojcik, J., Kikinis, R., Kubicki, M., 2014. Diffusion tensor imaging study of the fornix in first episode schizophrenia and in healthy controls. Schizophr. Res. 156 (2–3), 157–160. Goghari, V.M., Rehm, K., Carter, C.S., MacDonald 3rd, A.W., 2007. Regionally specific cortical thinning and gray matter abnormalities in the healthy relatives of schizophrenia patients. Cereb. Cortex 17 (2), 415–424. Goghari, V.M., Smith, G.N., Honer, W.G., Kopala, L.C., Thornton, A.E., Su, W., Macewan, G.W., Lang, D.J., 2013. Effects of eight weeks of atypical antipsychotic treatment on middle frontal thickness in drug-naive first-episode psychosis patients. Schizophr. Res. 149 (1–3), 149–155. Guo, W., Liu, F., Liu, Z., Gao, K., Xiao, C., Chen, H., Zhao, J., 2012. Right lateralized white matter abnormalities in first-episode, drug-naive paranoid schizophrenia. Neurosci. Lett. 531 (1), 5–9. Jbabdi, S., Johansen-Berg, H., 2011. Tractography: where do we go from here? Brain Connect. 1 (3), 169–183.
Jones, D.K., Knosche, T.R., Turner, R., 2013. White matter integrity, fiber count, and other fallacies: the do's and don'ts of diffusion MRI. Neuroimage 73, 239–254. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 13 (2), 261–276. Kendler, K.S., Lieberman, J.A., Walsh, D., 1989. The Structured Interview for Schizotypy (SIS): a preliminary report. Schizophr. Bull. 15 (4), 559–571. Knochel, C., O'Dwyer, L., Alves, G., Reinke, B., Magerkurth, J., Rotarska-Jagiela, A., Prvulovic, D., Hampel, H., Linden, D.E., Oertel-Knochel, V., 2012. Association between white matter fiber integrity and subclinical psychotic symptoms in schizophrenia patients and unaffected relatives. Schizophr. Res. 140 (1–3), 129–135. Kuroki, N., Kubicki, M., Nestor, P.G., Salisbury, D.F., Park, H.J., Levitt, J.J., Woolston, S., Frumin, M., Niznikiewicz, M., Westin, C.F., Maier, S.E., McCarley, R.W., Shenton, M.E., 2006. Fornix integrity and hippocampal volume in male schizophrenic patients. Biol. Psychiatry 60 (1), 22–31. Lebel, C., Gee, M., Camicioli, R., Wieler, M., Martin, W., Beaulieu, C., 2012. Diffusion tensor imaging of white matter tract evolution over the lifespan. Neuroimage 60 (1), 340–352. Leemans, A., Jeurissen, B., Sijbers, J., Jones, D.K., 2009. ExploreDTI: a graphical toolbox for processing, analyzing, and visualizing diffusion MR data. 17th Annual Meeting of Intl Soc Mag Reson Med, p. 3537. Sacher, J., Neumann, J., Okon-Singer, H., Gotowiec, S., Villringer, A., 2013. Sexual dimorphism in the human brain: evidence from neuroimaging. Magn. Reson. Imaging 31 (3), 366–375. Seidman, L.J., Pantelis, C., Keshavan, M.S., Faraone, S.V., Goldstein, J.M., Horton, N.J., Makris, N., Falkai, P., Caviness, V.S., Tsuang, M.T., 2003. A review and new report of medial temporal lobe dysfunction as a vulnerability indicator for schizophrenia: a magnetic resonance imaging morphometric family study of the parahippocampal gyrus. Schizophr. Bull. 29 (4), 803–830. Skudlarski, P., Schretlen, D.J., Thaker, G.K., Stevens, M.C., Keshavan, M.S., Sweeney, J.A., Tamminga, C.A., Clementz, B.A., O'Neil, K., Pearlson, G.D., 2013. Diffusion tensor imaging white matter endophenotypes in patients with schizophrenia or psychotic bipolar disorder and their relatives. Am. J. Psychiatry 170 (8), 886–898. Van Hecke, W., Leemans, A., D'Agostino, E., De Backer, S., Vandervliet, E., Parizel, P.M., Sijbers, J., 2007. Nonrigid coregistration of diffusion tensor images using a viscous fluid model and mutual information. IEEE Trans. Med. Imaging 26 (11), 1598–1612. Van Hecke, W., Sijbers, J., D'Agostino, E., Maes, F., De Backer, S., Vandervliet, E., Parizel, P.M., Leemans, A., 2008. On the construction of an inter-subject diffusion tensor magnetic resonance atlas of the healthy human brain. Neuroimage 43 (1), 69–80. Wechsler, D., 1999. Wechsler Abbreviated Scale of Intelligence. Pearson Education, Inc., San Antonio, TX.
