Introduction
It is recognized that individuals suffering from schizophrenia have a greater prevalence of the metabolic syndrome (De Hert, Dekker et al., Reference De Hert, Dekker, Wood, Kahl, Holt and Möller2009; De Hert, Schreurs, vancampfort, & Winkel, Reference De Hert, Schreurs, Vancampfort and Winkel2009; Mitchell et al., Reference Mitchell, Vancampfort, Sweers, Van Winkel, Yu and De Hert2013), and premature mortality (McGrath, Saha, Chant, & WElham, Reference McGrath, Saha, Chant and Welham2008; Saha & Chant, & McGrath, Reference Saha, Chant and McGrath2007) with a reduced life expectancy of approximately 20% compared with the general population (Hausswolff-Juhlin, Bjartveit, Lindström, & Jones, Reference Hausswolff-Juhlin, Bjartveit, Lindström and Jones2009; Newman & Bland, Reference Newman and Bland1991). The leading natural cause of death in patients with schizophrenia is cardiovascular disease (Brown, Kim, Mitchell, & Inskip, Reference Brown, Kim, Mitchell and Inskip2010; Hennekens, Reference Hennekens2006). However, the cause-effect relationships in this context is not straightforward (Hausswolff‐Juhlin et al., Reference Hausswolff-Juhlin, Bjartveit, Lindström and Jones2009). Genetics, lifestyle factors (e.g., high fat diets, alcohol, smoking, and lack of regular exercise), disease-specific symptoms (e.g., anhedonia, avolition, apathy), and pharmacological treatments (e.g., antipsychotic medications) may each play a role (De Hert, Schreurs, et al., Reference De Hert, Schreurs, Vancampfort and Winkel2009; De Hert, Detraux, Van Winkel, Yu, & Correll, Reference De Hert, Detraux, Van Winkel, Yu and Correll2012; Fritz-Wieacker et al., Reference Fritz-Wieacker, Matschinger, Heider, Schindler, Riedel-Heller and Angermeyer2007; Joukamaa et al., Reference Joukamaa, Heliövaara, Knekt, Aromaa, Raitasalo and Lehtinen2006; McCreadie, Reference McCreadie2003; Newcomer, Reference Newcomer2005).
Regarding modifiable lifestyle factors, 70–75% of patients with schizophrenia can be classified as physically inactive (Krogh, Speyer, Nørgaard, Moltke, & Nordeentoft, Reference Krogh, Speyer, Nørgaard, Moltke and Nordentoft2014; Lindamer et al., Reference Lindamer, McKibbin, Norman, Jordan, Harrison, Abeyesinhe and Patrick2008) and 70–86% are overweight or obese (Fritz-Wieacker et al., Reference Fritz-Wieacker, Matschinger, Heider, Schindler, Riedel-Heller and Angermeyer2007; Lee et al., Reference Lee, Hui, Chang, Chan, Li, Lee and Chen2013; McCreadie, Reference McCreadie2003). Compared to healthy individuals, patients with schizophrenia have reduced maximal oxygen uptake (VO2 max); therefore, they are at higher risk for cardiovascular disease contributing to a lower quality of life (Heggelund, Nilsberg, Hoff, Morke, & Helgerud, Reference Heggelund, Hoff, Helgerud, Nilsberg and Morken2011). VO2 max is measured by respiratory gas analysis during subjection to gradually increasing workloads to the point of exhaustion (Maud & Foster, 2006) during incremental exercise (i.e., progressively increasing work rate every n min) (Taylor, Buskirk, & Henschel, Reference Taylor, Buskirk and Henschel1955; Mitchell, Sproule, & Chapman, Reference Mitchell, Sproule and Chapman1958; Åstrand and Saltin, Reference Åstrand and Saltin1961. For review see Bassett & Howley, Reference Bassett and Howley2000). Peak oxygen uptake (VO2 peak) is sometimes used interchangeably with VO2 max for characterizing endurance during discontinuous exercise, where discrete square waves of different work rates are separated by periods of rest (Day, Rossiter, Coats, Skasick, & Whipp, Reference Day, Rossiter, Coats, Skasick and Whipp2003). A recent meta-analysis of the effects of physical activity in people with a severe mental illness indicated reductions in symptoms of schizophrenia and depressive symptoms, and improvements in anthropometric measures, aerobic capacity, and quality of life (Rosenbaum, Tiedemann, Sherrington, Curtis, & Ward, Reference Rosenbaum, Tiedemann, Sherrington, Curtis and Ward2014). Schizophrenia outpatient case studies have reported improvements in social interest and reductions in anxiety and depression with patients who exercise (Adams, Reference Adams1994; Pelham, Campaagna, Ritvo, & Birnie, Reference Pelham, Campagna, Ritvo and Birnie1993). There is also evidence that exercise therapy improves the cardiovascular fitness in patients with schizophrenia by increasing the VO2 max and peak work rate (Scheewe, Takken, Kahn, Cahn, & Backx, Reference Scheewe, Takken, Kahn, Cahn and Backx2012) while decreasing symptoms of schizophrenia, depression, and need of care, when compared with occupational therapy (Krogh et al., Reference Krogh, Speyer, Nørgaard, Moltke and Nordentoft2014; Scheewe, Backx, et al., 2013). High aerobic intensity training is reported to reduce distress and state anxiety, improve positive affect and well-being, enhance VO2 max and contribute to lower risk of cardiovascular disease (Heggelund, Kleppe, Morken, & Vedul-Kjelsås, Reference Heggelund, Kleppe, Morken and Vedul-Kjelsås2014); however, it had no effect on the psychiatric symptoms in patients with schizophrenia (Heggelund, Nilsberg, et al., Reference Heggelund, Hoff, Helgerud, Nilsberg and Morken2011). Other studies were able to show reductions of positive and negative symptoms in schizophrenia patients after aerobic exercise (Beebe et al., Reference Beebe, Tian, Morris, Goodwin, Allen and Kuldau2005; Pajonk et al., Reference Pajonk, Wobrock, Gruber, Scherk, Berner, Kaizl and Meyer2010), although the differences did not reach statistical significance due to small sample sizes.
Recent neuroimaging research has highlighted the effect of physical (in)activity on brain morphology in patients with schizophrenia. Pajonk et al. (Reference Pajonk, Wobrock, Gruber, Scherk, Berner, Kaizl and Meyer2010) showed a 12% hippocampal volume increase in the schizophrenia patients after aerobic exercise, in comparison with the non-exercising group, which was also related to improved short-term memory performance. Furthermore, the cognitive deficits in patients with schizophrenia, which affect attention, executive function, and memory (Malchow et al., Reference Malchow, Reich-Erkelenz, Oertel-Knöchel, Keller, Hasan, Schmitt and Falkai2013; Nuechterlein et al., Reference Nuechterlein, Barch, Gold, Goldberg, Green and Heaton2004), have been correlated with physical inactivity (Lee et al., Reference Lee, Hui, Chang, Chan, Li, Lee and Chen2013). Studies in healthy subjects have previously shown improvement in cognitive functioning, particularly visuospatial learning and memory processes with regular exercise (Colcombe & Kramer, Reference Colcombe and Kramer2003; Hillman, Erickson, & Kramer, Reference Hillman, Erickson and Kramer2008), which has been posited to be a consequence of neurogenesis in the adult hippocampal dentate gyrus which is driving the volume changes and associated improved functioning (Pereira et al., Reference Pereira, Huddleston, Brickman, Sosunov, Hen, McKhann and Small2007). Exercise-induced structural changes can also be due to increased growth factors, such as brain-derived neurotrophic factor (BDNF), which is highly expressed in the hippocampus, and essential in facilitating neurogenesis, synaptic plasticity, and memory and learning (Gomez‐Pinilla, Vaynman, & Ying, Reference Gomez‐Pinilla, Vaynman and Ying2008; Poo, Reference Poo2001). In chronic schizophrenia patients, improvements in physical fitness as a result of aerobic exercise training were related to increased serum BDNF levels (Kim et al., Reference Kim, Song, So, Lee, Song and Kim2014; Kimhy et al., Reference Kimhy, Vakhrusheva, Bartels, Armstrong, Ballon, Khan and Sloan2015). In a more recent study, Scheewe, Van Haren, et al. (2013) reported an association of cardiorespiratory fitness improvement with cortical thickening in the left frontal, temporal and cingulate cortices, and a reduction in lateral and third ventricle volume in patients with schizophrenia. Falkai et al. (Reference Falkai, Malchow, Wobrock, Gruber, Schmitt, Honer and Cannon2013), however, found no difference in cortical thickness in schizophrenia patients after aerobic exercise, although he reported increased right frontal and occipital cortical gray matter density in healthy subjects, suggesting that any exercise effects on cortical gray matter are likely to be attenuated in chronic schizophrenia. Mittal et al. (Reference Mittal, Gupta, Orr, Pelletier-Baldelli, Dean, Lunsford-Avery and Millman2013) studied ultra high-risk individuals for psychosis, who had 65% higher rates of sedentary behavior than age-matched controls and found a decrease in medial temporal volumes in the patient group compared to the healthy controls. Furthermore, the total level of physical activity in the patient group was correlated with higher gray matter volumes in bilateral parahippocampal gyri (Mittal et al., Reference Mittal, Gupta, Orr, Pelletier-Baldelli, Dean, Lunsford-Avery and Millman2013).
More research is needed to understand the effect and magnitude of physical activity on brain morphology in patients with schizophrenia, especially in the early phase of illness, to devise better preventive measures for cardiovascular disease and ways to improve cognitive functioning. The aim of this pilot study was to investigate the relationship between naturalistic physical activity and brain morphology in first-episode psychosis patients who have been classified as having high or low levels of regular physical activity. Physical activity levels were assessed using the International Physical Activity Questionnaire (IPAQ), which was used in schizophrenia patients in another study (Faulkner, Cohn, & Remington, Reference Faulkner, Cohn and Remington2006). We were interested in investigating whether there is a global and/or regionally specific difference in prefrontal and hippocampal volumes between the first-episode schizophrenia patients with high versus low physical activity levels, and whether we could find any relationship between physical activity, neurocognition, and brain morphometry between these two groups of patients. Our a priori hypothesis was that first-episode schizophrenia patients with high physical activity levels would have higher cortical thickness and brain volumes than first-episode schizophrenia patients with low physical activity levels. Since this was a novel, exploratory pilot investigation, we used one-tailed statistical testing to investigate the effect of physical activity for the first time in first-episode schizophrenia patients.
Method
Participants
Baseline data collected from 14 recent-onset, first-episode schizophrenia (FE Sz) patients are included in this pilot investigation. Participants were all enrolled in the UCLA Aftercare Research Program, which is a longitudinal, NIMH-funded outpatient clinical research program for schizophrenia patients with a recent first psychotic episode (i.e., within two years of onset). Patients were recruited from local inpatient and outpatient facilities in the Los Angeles area or are directly referred to the Program. Participants were provided with psychiatric medication management, psychoeducation, group skills training, and individual case management for 18 months as participants in the Program. Inclusion criteria were: (1) onset of a first psychotic episode within 24 months of program entry; (2) met criteria for schizophrenia, schizoaffective disorder, depressed type, or schizophreniform disorder (American Psychiatric Association, 1994); (3) age of 18 to 45 years; and (4) sufficient fluency in English to allow for valid completion of the testing protocol. DSM-IV diagnoses were made using the Structured Clinical Interview for DSM-IV (SCID). Symptom ratings were made using the 18-item Brief Psychiatric Rating Scale (BPRS) to assess positive, negative and affective symptoms over the last two weeks. Additionally, all FE Sz patients were prescribed risperidone, and baseline testing occurred after medication stabilization, typically within three months of outpatient program entry. The University’s institutional review board approved the protocol and all participants gave their written informed consent. Compliance with institutional research standards for animal or human research were completed in accordance with the Helsinki Declaration (http://www.wma.net/e/policy/ 17-c_e.html).
Participant information, which was collected by the clinician who conducted the diagnostic interview, included demographic data (age, gender, handedness, ethnicity, education, parental education). Trained research staff members collected the weight and height data using a standard hospital scale to calculate body mass index (BMI), and also administered and scored the neurocognitive battery (Nuechterlein et al., Reference Nuechterlein, Green, Kern, Baade, Barch, Cohen and Marder2008), the Short-Form International Physical Activity Questionnaire (IPAQ; Craig et al., Reference Craig, Marshall, Sjöström, Bauman, Booth, Ainsworth and Oja2003), and the Starting the Conversation dietary assessment questionnaire (STC; Paxton, Strycker, Toobert, Ammerman, & Glasgow, Reference Paxton, Strycker, Toobert, Ammerman and Glasgow2011). The STC is an eight-item food frequency instrument validated for use in primary care and health promotion settings to assess regular dietary patterns. The item scores are added to create a summary score (range, 0–16), with lower summary scores reflecting a healthier diet and higher scores reflecting a lower healthy quality diet. For the purposes of this pilot investigation we were interested in the group differences in symptom severity (e.g., BPRS total scores) and dietary habits (e.g., STC total score) to examine the effects of more severe symptoms or poorer dietary habits, which could be contributing to the differences in brain morphometry.
Physical Activity Assessment
Participants completed the interviewer administered IPAQ with the aid of a structured recall format that asked participants to recall activities for each of the last seven preceding days in morning, afternoon, and evening time periods. We used the 7-day reference period instead of asking what subjects “usually” did in a week given the preference for this format in the initial validation of the IPAQ (Craig et al., Reference Craig, Marshall, Sjöström, Bauman, Booth, Ainsworth and Oja2003).
Data from the IPAQ were summarized according to walking, moderate (e.g., carrying light loads, bicycling at a regular pace, or easy swimming), and vigorous activities (e.g., heavy lifting, digging, aerobics, or fast bicycling) per week. On the basis of what activities subjects self-reported, the interviewer also clarified the perceived intensity of that specific activity. Responses were converted to metabolic equivalents in minutes per week (METS-min/week) (Craig et al., Reference Craig, Marshall, Sjöström, Bauman, Booth, Ainsworth and Oja2003). According to the IPAQ scoring protocol: total minutes over last seven days spent on vigorous activity, moderate-intensity activity, and walking were multiplied by 8.0, 4.0, and 3.3, respectively, to create MET scores for each activity level. The low physical activity group met the following criterion, as indicated by the IPAQ scoring protocol: (1) No activity or (2) Less than: (a) 3 days of vigorous-intensity activity of at least 20 min/day or (b) 5 days of moderate-intensity activity and/or walking of at least 30 min/day or (c) 5 days of any combination of walking, moderate-intensity or vigorous intensity activities with minimum total physical activity of at least 600 MET min/week. The high physical activity group met the following criterion: (1) vigorous-intensity activity on at least 3 days minimum total physical activity of at least 1500 MET min/week or (2) 7 or more days of any combination of walking, moderate-intensity or vigorous-intensity minimum total physical activity of at least 3000 MET min/week.
Neurocognitive Assessment
The Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) Consensus Cognitive Battery (MCCB) (Nuechterlein et al., Reference Nuechterlein, Green, Kern, Baade, Barch, Cohen and Marder2008) includes seven cognitive domains: speed of processing (Tests: Category Fluency, BACS Symbol coding, Trail Making A), attention/vigilance (Continuous Performance-Identical Pairs Version), working memory (Letter Number Span, WMS-II Spatial Span), verbal learning (Hopkins Verbal Learning Test-R), visual learning (Brief Visuospatial Memory Test-R), reasoning and problem solving (NAB Mazes), and social cognition (MSCEIT Managing Emotions). In addition to the seven domain scores, the MCCB also provides an overall composite score, an index of cognitive functioning across domains. The overall composite score is derived through equal weighting of the seven MCCB domain scores (Nuechterlein et al., Reference Nuechterlein, Green, Kern, Baade, Barch, Cohen and Marder2008). All analyses of MCCB scores used age and gender correction based on the MCCB computer scoring program. Based on previous studies of the effects of physical activity on cognition in schizophrenia (Malchow et al., Reference Malchow, Keller, Hasan, Dörfler, Schneider-Axmann, Hillmer-Vogel and Falkai2015; Pajonk et al., Reference Pajonk, Wobrock, Gruber, Scherk, Berner, Kaizl and Meyer2010) we looked at group differences from the sub-domain of verbal learning, in addition to the overall MCCB composite score. We also conducted an exploratory investigation of social cognition, due to the higher levels of social abilities seen in clinical samples of patients who regularly exercise (Taylor, Sallis et al., Reference Taylor, Sallis and Needle1985).
Neurocognitive testing procedures for administration of the MCCB have been previously described (Nuechterlein et al., Reference Nuechterlein, Green, Kern, Baade, Barch, Cohen and Marder2008). Briefly, FE Sz patients completed the MCCB as part of their baseline assessment. Raw scores for each MCCB test were converted to age and gender corrected t scores for the general population using the MCCB scoring program (Mean=50; SD=10). For all participants, testing was conducted by bachelor’s level examiners who received extensive training in administration of the MCCB. All examiners participated in periodic checks on MCCB administration and scoring practices.
Brain Imaging
MRI image acquisition
All scanning was performed on a Siemens Trio 3 Tesla (T) scanner with a 12-channel head coil at the UCLA Staglin Center for Cognitive Neuroscience. Subject’s head movement was minimized with the use of foam padding. A high-resolution T1-weighted anatomical scan was acquired using a magnetization-prepared rapid acquisition gradient echo (MPRAGE) sequence with a repetition time (TR)=2300 ms, echo time (TE)=2.91 ms, inversion time (TI)=900 ms, flip angle=9 degrees, and field-of-view (FOV) of 256 mm (anterior-to-posterior)×240 mm (superior-to-inferior)×176 mm (left-to-right), resulting in a voxel resolution of approximately 1 mm×1 mm×1.2 mm.
Segmentation, volumetry cortical reconstruction, and cortical thickness calculation
Image processing and analyses of MR data including brain segmentation, cortical reconstruction and cortical thickness estimations were conducted in the UCLA Brain Tumor Imaging Laboratory (BTIL) within the Center for Computer Vision and Imaging Biomarkers (CVIB). MPRAGE scans were first assessed for quality to ensure the absence of artifacts, such as aliasing. The MR data were then processed via FreeSurfer (version 4.3; http://surfer.nmr.mgh.harvard.edu) cortical reconstruction pipeline wherein each subjects cortical surface and thickness at each vertex were computed using a semi-automated approach previously described in detail (Dale & Sereno, Reference Dale and Sereno1993; Dale, Fischl, & Sereno, Reference Dale, Fischl and Sereno1999, Fischl, Sereno, & Dale, Reference Fischl, Sereno and Dale1999, Fischl, Sereno, Tootell, & Dale Reference Fischl, Sereno and Dale1999; Fischl & Dale, Reference Fischl and Dale2000; Salat et al., Reference Salat, Buckner, Snyder, Greve, Desikan, Busa and Fischl2004). In short, automated serial manipulations of MR data for cortical rendering included: (1) transforming the 3D, T1-weighted MRI data into Talairach coordinates; (2) normalizing image signal intensity to correct for unwanted variations in intensity due to RF-field inhomogeneity; (3) the stripping of the skull and other extra-cerebral tissue using a watershed algorithm; (4) parcellating and labeling the white matter volume based upon normalized intensity; (5) correcting topological errors and smoothing the generated surfaces; and (6) the construction of cortical surface from the white/gray matter interface to the pial surface at the gray matter/CSF interface. The resulting segmentations were visually inspected on a slice-by-slice basis to ensure correct delineation of pial from dura surfaces and parcellation of subcortical white matter structures. Manual editing of the pial surface and white matter was done as needed to improve the accuracy of the segmentation. Cortical thickness measurements were then obtained from calculating the distance between pial surface and the gray/white matter boundary (Dale and Sereno, Reference Dale and Sereno1993; Dale et al., Reference Dale, Fischl and Sereno1999). The volume of gray matter and subcortical white matter structures including the hippocampus were measured automatically using FreeSurfer as described elsewhere (Fischl et al., Reference Fischl, Salat, Busa, Albert, Dieterich, Haselgrove and Dale2002). Following this procedure, resulting three dimensional cortical surfaces were aligned to a standardized mesh surface with a mesh density linear depth of 141 for use in group comparisons, blurred using a Gaussian filter with a full width at half max (FWHM) of 8 mm, then labels of the region of interest comprising the prefrontal cortex (PFC) and hippocampus were examined in standard space and gray and white volumes were tabulated. Additionally, in lieu of a false discovery rate analysis, due to the limited sample size and degrees of freedom, we chose to use interactive clustering on the surface data to achieve the critical cluster size corresponding to a significance value of p≤.05. The application of the cluster size area 100 mm2 is therefore meant to reflect the corrected p-value of.05. Lastly, comparisons between patient groups were performed and cortical thickness comparisons were visualized in standard space.
Statistical Analysis
All analyses were performed in the statistical software SPSS 22.0. Before conducting group analyses we conducted Shapiro-Wilk’s tests to asses the normal distribution in this small sample. From the tests, we found that brain volumes, cognition, and physical data were all not significant, indicating normality and we concluded that the parametric t tests and Pearson correlations were the appropriate tests. All demographic, clinical, physical, and neurocognitive data were compared across groups using Student’s independent t tests (two-tailed α=.05). Group comparisons of brain volumes and cortical thickness measurements from a priori regions of interest, including total cortical gray matter, PFC, and hippocampus (Hillman et al., Reference Hillman, Erickson and Kramer2008; Mittal et al., Reference Mittal, Gupta, Orr, Pelletier-Baldelli, Dean, Lunsford-Avery and Millman2013; Pajonk et al., Reference Pajonk, Wobrock, Gruber, Scherk, Berner, Kaizl and Meyer2010; Scheewe et al., Reference Scheewe, Takken, Kahn, Cahn and Backx2012) were analyzed using Student’s independent t tests (one-tailed α=.05), since we had a specific hypothesis on directionality in a sample of FE Sz patients. Also due to the small sample size and exploratory nature of this investigation Cohen’s d effect size estimates were calculated for brain volume, cortical thickness and cognitive performance data.
As an exploratory investigation, we sought to test for associations between brain volume/cortical thickness and physical measurements (e.g., BMI) and also between physical activity levels (e.g., Total METS/week) and cognition (MCCB: Verbal learning, social cognition and overall composite MCCB scores). Pearson bivariate correlations were carried out across groups to explore potential relationships (two-tailed α=.05).
Results
Demographic and Clinical Characteristics
Demographic, clinical, and physical data for all participants can be found in Table 1. There were no significant differences in age, sex, handedness, ethnicity, patient or parental education, or symptom ratings. All patients were on oral risperidone awaiting random assignment to treatment with either oral or injectable risperidone as part of the research protocol. Physical measurements showed that there was not a significant difference in BMI, even though the high physical activity group had a significantly higher amount of METS min/week (t=21.99; p=.001). Additionally, no group differences were found between dietary habits derived from the STC questionnaire.
Table 1 Demographic, clinical, physical measurements, and neurocognitive data of first-episode schizophrenia patients, grouped by physical activity levels
a Significance tests were two-tailed.
b Age was defined as age at MRI scanning.
PA=physical activity; BPRS=Brief Psychiatric Rating Scale; BMI=body mass index; METS=metabolic equivalents; IPAQ=International Physical Activity Questionnaire; STC=Starting the Conversation; MCCB=MATRICS Consensus Cognitive Battery.
Group Differences in Cognitive Performance
Baseline cognitive data are presented in Table 1. Although there was not a significant difference between groups on the neuropsychological measures, there was a non-significant tendency toward lower cognitive functioning on all three measures in the low physical activity group. The largest effect size observed was for social cognition (Cohen’s d=1.43), a moderate effect size for verbal memory (d=0.50), and a small effect size for the MCCB overall composite (d=0.32).
Group Differences in Brain Volumes
Comparisons of total cortical gray matter volume, total PFC gray matter volume, left PFC gray matter volume, right PFC gray matter volume all yield significantly reduced volumes in the low physical activity group (p=.03; Cohen’s d=1.09; p=.04, d=1.00; p=.04, d=1.02; and p=.05, d=0.98, respectively) (Table 2). When we examined the cortical parcel sub-regions compromising the PFC, we found that although almost all sub-regions were reduced in the low physical activity group eight focal regions were potentially driving the differences in overall PFC and included: left middle frontal gyrus (p=.05), left orbital gyrus (p=.02), left precentral gyrus (p=.04), left superior frontal sulcus (p=.05), left superior part of the precentral sulcus (p=.05), right traverse frontopolar gyrus and sulcus (p=.04), right straight gyrus (p=.005), and right inferior part of the precentral sulcus (p=.01). Figure 1 displays a hippocampal statistical distribution for the two patient groups. There was also a significant reduction in total hippocampal volume (p=.05; d=0.94) and left hippocampal volume (p=.05; d=0.95) in the low physical activity group (Table 2), but not for the right hippocampal volume, although it was on average smaller in the low activity group (p=.07; d=0.84).
Fig. 1 Hippocampal probability distribution region of interest overlaid atop of MNI atlas brain. (a, b) Patient brains and hippocampal ROI data were registered to MNI space and then averaged, resulting in high physical activity patient population hippocampal probability distribution and low physical activity patient population hippocampal probability distribution. (c, d) Magnified High PA (top) and Low PA (bottom) probability distributions to underscore the difference in mean hippocampal volume between the two groups. PA=physical activity.
Table 2 Baseline volumetry and group differences for total cortical gray matter volume and regions of interest
a Significance tests were one-tailed.
PFC=Prefrontal Cortex.
Group Differences in Cortical Thickness
Figure 2 displays the average cortical thickness from each group. Figure 3 shows a corrected statistical map threshold of p=.05 for whole brain cortical thickness, showing regions that had reduced thickness in the low physical activity group compared to the high physical activity group, overlaid onto a standardized brain. Cortical thickness within sub-regions of the PFC was significantly reduced in the low physical activity group in the dorsolateral and orbitofrontal cortex. Table 3 lists the location, anatomic labels, and comparison statistics for significant clusters.
Fig. 2 Standardized 3D cortical renders (left panel) and inflated cortex (right panel) overlaid with the average cortical thickness between the high physical activity group (a) and low physical activity group (b). PA=physical activity.
Fig. 3 Cortical thickness t test comparison of high and low physical activity groups overlaid on standardize surface mesh displaying brain regions of greater cortical thickness. The t-statistical map is thresholded to significance value of p=.05 and a cluster area threshold of 100 mm2. Regions highlighted in the blue circle signify brain regions in the dorsolateral and orbitofrontal prefrontal cortex which had reduced cortical thickness in the low physical activity group compared to the high physical activity group.
Table 3 Location, anatomic labels, and comparison statistics for significant clusters from the cortical thickness analysis
PA=physical activity.
Correlations of Brain Volumes, Cortical Thickness, and Cognition
In an exploratory analysis we examined the relationship between brain volume/cortical thickness and physical measurements (e.g., BMI) and also between physical activity levels (e.g., Total METS/week) and neurocognition (MCCB: Verbal memory, social cognition and overall composite scores). Although none of the correlations were significant, there was a non-significant tendency toward relationships between the amount of total METS min/week (e.g., total physical activity) and performance on verbal learning tests (r=.49; p=.08) and social cognition tests (r=.52; p=.06) (Figure 4). There was also a non-significant tendency between lower BMI and larger left hippocampal volume (r=−.50; p=.09) (Figure 4).
Fig. 4 Correlations in the combined sample of first-episode schizophrenia patients (n=14). Scatterplots showing the near significant relationship between METS min/week (e.g., Total physical activity per week) and neurocognitive performance and hippocampal volume and BMI. Trend level correlations between the amount of total METS min/week (e.g. total physical activity) and performance scores in (a) verbal learning and memory (r=.49; p=.08) and (b) social cognition (r=.52; p=.06). (c) Larger left hippocampal volume also showed a trend for a correlation with lower BMI (r=−.50; p=.09). METS=metabolic equivalents; BMI=body mass index.
Discussion
To our knowledge, this is the first study to investigate the role of naturalistic physical activity levels on brain structure and cortical thickness in first-episode schizophrenia. This investigation is an important starting point in a burgeoning field of studying the role of exercise as a novel intervention to increase brain plasticity and improve cognition in schizophrenia. We found that FE Sz patients with low self-reported levels of physical activity had reductions in total gray matter volume, specifically within the PFC and hippocampus. We also found a reduction in cortical thickness in the PFC regions of the dorsolateral and obitofrontal cortex in patients with lower physical activity levels. In an exploratory analysis, we found non-significant tendencies toward relationships in FE Sz patients between more physical activity and cognitive performance on tests of verbal learning and social cognition. Lastly, we found a trend level correlation between lower BMI and larger left hippocampal volume. Together these findings highlight the possibility that physical activity has an potential impact on brain morphometry in FE Sz patients and supports the need to initiate interventions promoting physical exercise in this patient population not only to stave off deletrious physical health but also to improve cognitive functioning and brain health. This study also highlights that before starting an exercise intervention study, researchers need to examine the role of baseline levels of physical activity and their relationships to brain structures at study entry, as this may have an effect on the magnitude of the treatment response.
Previous research on the benefits of physical activity on brain plasticity in older sedentary adults found that increasing physical exercise increased cerebral gray matter volume (Colcombe et al., Reference Colcombe, Erickson, Scalf, Kim, Prakash, McAuley and Kramer2006), and more specifically within the PFC (Erickson, Miller, & Roecklein, Reference Erickson, Miller and Roecklein2012), suggesting the role of increasing physical activity on overall brain health and structural integrity. In line with our PFC findings, a recent study with chronic schizophrenia patients, Scheewe et al. (2013) reported an association between cardiorespiratory fitness improvement with cortical thickening in the left frontal, temporal and cingulate cortices, and a reduction in lateral and third ventricle volume in patients with schizophrenia. Falkai et al. (Reference Falkai, Malchow, Wobrock, Gruber, Schmitt, Honer and Cannon2013), however, found no difference in cortical thickness in schizophrenia patients after aerobic exercise, although they did find increased right frontal and occipital cortical gray matter density in healthy subjects, suggesting that any exercise effects on cortical gray matter may be attenuated in chronic schizophrenia.
The hippocampus is a brain region most commonly associated with the greatest sensitivity to volume increases as a result of higher physical activity levels. Much of this work validating this relationship between increased exercise, hippocampal size, and memory functioning has been carried out in mice (van Praag, Christie, Sejnowski, & Gage, Reference Van Praag, Christie, Sejnowski and Gage1999) and aging humans (Erickson et al., Reference Erickson, Miller and Roecklein2012), but also in chronic schizophrenia patients (Pajonk et al., Reference Pajonk, Wobrock, Gruber, Scherk, Berner, Kaizl and Meyer2010) and ultra-high risk for psychosis youth (Mittal et al., Reference Mittal, Gupta, Orr, Pelletier-Baldelli, Dean, Lunsford-Avery and Millman2013). Research from Mittal et al. (Reference Mittal, Gupta, Orr, Pelletier-Baldelli, Dean, Lunsford-Avery and Millman2013) in youth at high-risk for developing psychosis revealed that these patients at an imminent risk for developing schizophrenia have lower objectively recorded physical activity levels than matched healthy controls. Furthermore, they found that those at high-risk who also have a more sedentary lifestyle had smaller medial temporal structure volumes. Although they were not able to provide data on conversion to psychosis in the high-risk sample, this study highlights that lower physical activity levels and related smaller medial temporal lobe volumes might be potential treatment targets to delay or prevent conversion to a full-blown psychotic illness. These exercise-induced structural changes in the hippocampus have been found to be related to increased growth factors such as BDNF, which are highly expressed in the hippocampus, and essential for facilitating neurogenesis, synaptic plasticity, memory, and learning (Gomez‐Pinilla et al., Reference Gomez‐Pinilla, Vaynman and Ying2008; Poo, Reference Poo2001), which could serve as a target mechanism in exercise interventions.
Our findings suggest that higher amounts of physical activity levels are associated with larger total brain volumes and in the PFC and hippocampus, along with cortical thickness in the PFC in FE Sz patients. Although the underlying mechanisms of brain volume increases as a result of increased physical fitness is still relatively speculative in humans, increased production of neurotrophins, (e.g., BDNF, insulin-like growth factor-1, vascular endothelial growth factor), increased neurogenesis, syntaptogenesis, vascularization, and improved cell energy metabolism all seem to play a critical role in animal studies (Cotman, Berchtold, & Christie, Reference Cotman, Berchtold and Christie2007; Neeper, Gómez-Pinilla, Choi, & Cotman, Reference Neeper, Gómez-Pinilla, Choi and Cotman1995; van Praag, Reference Van Praag2008). In schizophrenia patients, there have been two published studies that have explored the role of serum-levels of BDNF before and after an exercise intervention (Kim et al., Reference Kim, Song, So, Lee, Song and Kim2014; Kimhy et al., Reference Kimhy, Vakhrusheva, Bartels, Armstrong, Ballon, Khan and Sloan2015). Both studies found elevated serum levels change in BDNF from baseline to 12-week follow-up, and one study also found a relationship between improved physical fitness and cognition, measured by the MCCB composite score (Kimhy et al., Reference Kimhy, Vakhrusheva, Bartels, Armstrong, Ballon, Khan and Sloan2015).
Although the present correlations of physical activity levels and cognitive functioning were not significant, there was a large effect size with social cognition and a medium effect size with verbal learning. These findings were further supported by the non-significant tendency toward a correlation between total amounts of physical activity completed per week and higher scores on these neurocognitive measures. The finding of greater physical activity and increased verbal memory abilities is in line with intervention studies of exercise in chronic schizophrenia patients (Malchow et al., Reference Malchow, Keller, Hasan, Dörfler, Schneider-Axmann, Hillmer-Vogel and Falkai2015; Pajonk et al., Reference Pajonk, Wobrock, Gruber, Scherk, Berner, Kaizl and Meyer2010). The finding of a tendency for higher social cognitive functioning to be associated with greater physical activity is a novel finding. Previous research has suggested that exercise can improve symptoms of depression, social, anhedonia and low self-esteem in patients with schizophrenia, which may provide the link between greater amounts of physical activity and social cognitive functioning (Gorczynski & Faulkner Reference Gorczynski and Faulkner2010). Additionally, one study in schizophrenia spectrum disorder patients found a relationship between greater amounts of sedentary behavior and lower performance on a test of metacognition (Snethen, McCormick, & Lysaker, Reference Snethen, McCormick and Lysaker2014). We would posit that the increase in higher order social cognition is supported by the underlying increased PFC volume in the higher activity patients, which subserves these processes.
Although there is a growing literature to support the need to develop effective interventions to promote a physically active lifestyle in chronic schizophrenia patients, to date, there has been only one published exercise intervention study in FE Sz patients (Abdel-Baki, Brazzini-Poisson, Marois, Letendre, & Karelis, Reference Abdel-Baki, Brazzini-Poisson, Marois, Letendre and Karelis2013). Although it was not a randomized controlled trial, it did show the feasibility of a 14-week aerobic exercise intervention and the ability of the intervention to improve physical fitness in a FE Sz sample. There have yet to be any published studies of the effects of exercise on cognition and brain structure in FE Sz. Published exercise interventions in schizophrenia samples, therefore, are limited to chronic patient samples (illness duration/chronicity ~10 years). These aerobic exercise interventions range in duration from 6–24 weeks, requiring ~90 min of exercise per week, have high adherence to supervised group exercise training (79%) and show significant improvements in physical health and symptomatology (for review, see Firth, Cotter, Elliott, French, & Yung, Reference Firth, Cotter, Elliott, French and Yung2015). Together these studies show the feasibility of intensive exercise interventions in schizophrenia patients and the ability to reduce deleterious health outcomes and improve functioning. The next step is to develop small randomized controlled exercise trials to test feasibility and target behavioral outcomes in FE Sz. Then larger treatment trials can be implemented to explore the neural mechanisms underlying improvements in brain health and cognitive functioning in FE Sz.
Our study results must be evaluated in the context of some limitations. The sample size of this pilot investigation is very small. Further research of physical activity levels in larger samples is warranted to determine the relationship between brain morphometry and physical activity levels in FE Sz patients. Also due to the small sample size for the volumetric imaging analysis, we reported one-tailed significance tests and effect sizes, which proved to be a useful method for our pilot and exploratory analysis but need to be replicated with greater statistical rigor in a larger sample. We did seek to explore potential confounds that would affect brain structures that were not related to activity levels, such as anti-psychotic medication type (which was the same for all patients in our sample), age, gender, education, BMI, symptom severity, and dietary habits, and we did not find any group differences. Future studies should also take into account these potentially confounding factors. Additionally, our study groupings depended on a self-report tool for the assessment of physical activity habits, which may influence the validity and reliability of recalled data, even though the IPAQ questionnaire used in our study had been previously used in another study with schizophrenia patients (Faulkner et al., Reference Faulkner, Cohn and Remington2006). Concern regarding the inconsistency between self-reported and directly measured physical activity were recently described by Prince et al. (Reference Prince, Adamo, Hamel, Hardt, Gorber and Tremblay2008), who systematically reviewed the literature to determine the extent of discrepancy between subjectively (self-report, e.g., questionnaire, diary) and objectively (directly measured; e.g., accelerometry) assessed adult physical activity. Their findings suggest that self-report measures were both higher and lower than directly measured levels of physical activity, although no clear trends emerged in the over- or underreporting of physical activity by self-report compared to direct methods (Prince et al., Reference Prince, Adamo, Hamel, Hardt, Gorber and Tremblay2008). Because of the potential over-estimation of the physical activity values by subjective measures (Sebastiao et al., Reference Sebastião, Gobbi, Chodzko-Zajko, Schwingel, Papini, Nakamura and Kokubun2012), in future studies we would also seek to collect objective longitudinal measures of current physical activity habits, using actigraphs or pedometers. However, these measurements also have their limitations since they could interfere with naturalistic data collection and cause a change in the patients’ typical physical activity behaviors. Objective measures of physical fitness, such as VO2 max, would help to assess cardiovascular fitness to determine group classification (e.g., low vs. high fitness levels). It is also likely that the correlation between physical activity and social cognition and/or verbal learning might be due to the contribution of several mechanisms rather than a single mechanism acting in isolation (Faulkner & Carless, Reference Faulkner and Carless2006). Therefore, we would also like to collect more historical data on lifetime physical activity habits, as current activity levels after having been diagnosed with psychosis may not be reflective of activity levels before illness onset. We currently have an randomized controlled trial under way to examine the effects of a physical activity intervention on brain structure, neurotrophin levels, and cognitive functioning in FE Sz patients.
Acknowledgments
We thank the participants for giving their time to this study. We gratefully acknowledge the assistance of the UCLA Aftercare Research Program research assistants Jacqueline Hayata, B.A., and Livon Ghermezi, B.A., and clinic staff Laurie Casaus, M.D., John Luo, M.D., Yurika Sturdevant, Psy.D., and Luana Turner, Psy.D. None of the study sponsors had any role in the data collection, analysis, or write up of the current paper. The authors have declared that there are no conflicts of interest in relation to the subject of this study. This work was supported by the National Institute of Mental Health (S.M., K01 MH099431), (K.N., R34 MH102529), (K.N., P50 MH066286).
