|Home | About | Journals | Submit | Contact Us | Français|
Enlarged ventricles and reduced hippocampal volume are consistently found in patients with first-episode schizophrenia. Studies investigating brain structure in antipsychotic-naive patients have generally focused on the striatum. In this study, we examined whether ventricular enlargement and hippocampal and caudate volume reductions are morphological traits of antipsychotic-naive first-episode schizophrenia.
We obtained high-resolution 3-dimensional T1-weighted magnetic resonance imaging scans for 38 antipsychotic-naive first-episode schizophrenia patients and 43 matched healthy controls by use of a 3-T scanner. We warped the brain images to each other by use of a high-dimensional intersubject registration algorithm. We performed voxel-wise group comparisons with permutation tests. We performed small volume correction for the hippocampus, caudate and ventricles by use of a false discovery rate correction (p < 0.05) to control for multiple comparisons. We derived and analyzed estimates of brain structure volumes. We grouped patients as those with (n = 9) or without (n = 29) any lifetime substance abuse to examine the possible effects of substance abuse.
We found that hippocampal and caudate volumes were decreased in patients with first-episode schizophrenia. We found no ventricular enlargement, differences in global volume or significant associations between tissue volume and duration of untreated illness or psycho-pathology. The hippocampal volume reductions appeared to be influenced by a history of substance abuse. Exploratory analyses indicated reduced volume of the nucleus accumbens in patients with first-episode schizophrenia.
This study was not a priori designed to test for differences between schizophrenia patients with or without lifetime substance abuse, and this subgroup was small.
Reductions in hippocampal and caudate volume may constitute morphological traits in antipsychotic-naive first-episode schizophrenia patients. However, the clinical implications of these findings are unclear. Moreover, past substance abuse may accentuate hippocampal volume reduction. Magnetic resonance imaging studies addressing the potential effects of substance abuse in antipsychotic-naive first-episode schizophrenia patients are warranted.
Magnetic resonance imaging (MRI) studies have demonstrated the presence of structural brain abnormalities in multiple brain regions in chronic schizophrenia patients compared with healthy controls.1 Although, volume changes have also been observed in various brain regions in first-episode schizophrenia patients,2 only hippocampal volume reduction and ventricular enlargement are consistently present, as shown in 2 recent meta-analyses.3,4 Inconsistencies among studies are probably because of differences in methods, samples sizes and sample composition (e.g., studies vary in the inclusion criteria regarding previous exposure to anti-psychotic medications and substance abuse or dependence).
Numerous studies point toward hippocampal involvement in schizophrenia (for reviews see5,6). Volumetric reductions in hippocampus, however, are not pathognomonic for schizophrenia, but they have been associated with numerous neuropsychiatric disorders, including substance abuse.7 Moreover, typical and atypical antipsychotics may affect hippocampal volumes differently.8 In antipsychotic-naive patients, significant hippocampal reductions have been reported,9 although not consistently.10 The clinical implications of hippocampal involvement in antipsychotic-naive schizophrenia patients are unclear. However, hippocampal reductions may be associated with illness duration and psycho pathology.11
Ventricular enlargement is also not specific for schizophrenia.12 Nevertheless, recent studies suggest that ventricular enlargement may progress during the course of the disease and may be related to outcome13 or antipsychotic treatment.14
Studies investigating minimally medicated and antipsychotic-naive schizophrenia patients have often focused on the striatum, which consists of the caudate nucleus, putamen and nucleus accumbens.15 Striatal volume reductions are not consistently found in medicated first-episode patients, likely because antipsychotic treatment in itself can induce alterations in the striatum (for review see16). Nevertheless, a recent meta-analysis of voxel-based structural MRI studies in first-episode patients found the presence of caudate nucleus reductions as compared with chronic schizophrenia patients.2 Moreover, nonpsychotic children of psychotic individuals have been shown to have smaller caudate nuclei.17 Absolute volume reductions in the putamen and nucleus accumbens15,18,19 have also been observed, although significant changes have only been observed in the putamen.18 Significantly increased putamen volumes have also been reported.9
Although the clinical implications of structural striatal changes in schizophrenia are unresolved, caudate volume changes have been associated with illness duration and positive symptoms.20 Also, a positive correlation was found between putamen surface contractions and affective flattening.18
Structural changes already observed in antipsychotic-naive schizophrenia patients are likely not because of hospitalization, chronicity or antipsychotic treatment. However, schizophrenia patients commonly have a past or present history of substance abuse or dependence. Both alcoholism21,22 and cannabis abuse23,24 have been associated with grey matter changes in frontal, temporal and subcortical regions as well as ventricular changes. Some studies have dealt with this issue by excluding patients with comorbid substance abuse or dependence.25–27 However, the criteria used are not uniform among studies and encompass exclusion because of substance dependence20, “significant abuse”19 or abuse within a 6-month period,28 as well as no mention of abuse.9,10,29
The primary aim of our study was to investigate whether the presence of hippocampal reduction, ventricular enlargement and caudate reduction is a morphologic trait in first-episode antipsychotic-naive schizophrenia patients compared with matched healthy controls. Regional voxel-wise and volumetric analyses were performed after high-dimensional intersubject warping of all patients’ brains.30 Region-of-interest (ROI) masks of the hippocampus, ventricles and caudate were created to test our a priori hypotheses. Moreover, associations with clinical measures were explored. Finally, we tested for possible effects of any lifetime substance abuse.
The study was conducted in accordance with the declaration of Helsinki II and approved by the ethics committee of the Capital Region (H-KF-01–78/97). After complete description of the study to the participants, written informed consent was obtained.
Initially, 43 patients and 43 healthy controls, matched for age, sex and parental socio-economic status, underwent MRI scans. Patients were recruited as part of a first-episode schizophrenia study conducted in the Capital Region of Copenhagen, Denmark (Psychiatric Centres Amager, Ballerup, Bispebjerg, Gentofte, Glostrup and Rigshospitalet). We included patients aged 18–45 years with a diagnosis of schizophrenia, no prior exposure to antipsychotic medication and no medical or neurologic comorbidity. The DSM-IV diagnoses were based on the Schedules for Clinical Assessment in Neuropsychiatry (SCAN), version 2.1.31 We also included patients who used benzodiazepines (to reduce agitation and anxiety). Use of antidepressants was recorded. Patients with any lifetime substance abuse are denoted Ptab, patients with no lifetime substance abuse diagnosis are denoted Ptnon-ab and the total patient sample is denoted Ptall.
The controls were recruited from the community and had no prior or present psychiatric disorder, had never used psychotropic medication and had no first-degree relatives with a psychiatric disorder, as determined by SCAN interviews. Both patients and controls had normal physical and neurologic examinations, no history of major head injury (loss of consciousness), no mental retardation, no contraindications on MRI or any nonpsychiatric disorder. We assessed handedness by use of the Edinburgh Inventory.32 A neuroradiologist (A.L.) examined the MRI scans, which were free of pathology. We excluded controls with substance abuse or dependence.
Trained raters assessed psychopathology with the Positive and Negative Syndrome Scale (PANSS),33 and the interviews were recorded on DVDs for validation purposes. In a random subset of 10 PANSS recordings, an intraclass correlation of 0.92 in a 2-way mixed effect model was achieved.
We defined the duration of untreated illness as the time between the first unspecific symptoms related to psychosis to the date of the MRI scan. Symptoms had to be associated with a decline in a previous stable level of function. We collected data about the duration of untreated illness with the best-estimate approach25 with information from the SCAN interview, clinical records and relatives, if possible.
We acquired high-resolution 3-dimensional (3-D) T1-weighted, sagittal, magnetization-prepared rapid-gradient echo (MPRAGE) scans of each patient’s whole head (echo time 3.93 ms, repetition time 1540 ms, inversion time 800 ms, flip angle 9°, field of view 256 mm, matrix 256 × 256, 1 × 1 × 1 mm voxels, 192 slices) and 2-dimensional (2-D) T2-weighted, axial, turbo spin echo (TSE) scans of the whole brain (echo time one 17 ms, echo time two 100 ms, repetition time 9000 ms, flip angle 150°, field of view 220 mm, matrix 256 × 256, GRAPPA acceleration factor 2, 30 reference lines, 0.9 × 0.9 × 3 mm voxels, 50 slices). We used a Siemens Magnetom Trio 3-T scanner with an 8-channel head coil (Invivo Corporation).
We corrected the images for spatial distortions owing to nonlinearity in the gradient system of the scanner34 and processed the images using the VBM5 toolbox (http://dbm.neuro.uni-jena.de/vbm/vbm5-forspm5/) in SPM5 (Well-come Department of Cognitive Neurology, University College London, UK), which includes a unified segmentation algorithm,35 and a hidden Markov random field method.36 We used the T2-weighted images to automatically create brain masks in native space. We derived brain masked grey and white and cerebral spinal fluid tissue maps in native space from the T1 images. We used these masks, together with the affine part of the spatial transformation from native to Montreal Neurological Institute (MNI) space, in Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra (DARTEL)30 using default settings, allowing for high-dimensional intersubject registration. Recently, it has been shown that DARTEL can successfully register the hippocampus across patients.37 Using the final flow fields that parameterize the deformations, brain masked grey matter, white matter and cerebral spinal fluid images were warped into average image space (DARTEL space) and modulated with the Jacobian determinant of the applied deformation fields to correct for local volume changes following the high dimensional intersubject warping. We used voxel-wise analyses to test for differences in regional tissue volume. We smoothed these tissue images with an 8-mm full-width at half-maximum Gaussian kernel. We visually checked the quality of the images generated at each processing stage.
We created the following ROI masks on the average of DARTEL warped MPRAGE images for all participants. The delineation landmarks were as follows: hippocampus,38 ventricles, lateral and third ventricle,39 caudate nucleus, nucleus accumbens and putamen.15
We acquired intracranial volume estimates by integrating and adding image intensity values of modulated and warped grey matter, white matter and cerebral spinal fluid images. Total brain volume was acquired by integrating and adding grey matter and white matter image intensity values. Total ROI volumes were acquired to investigate whether local voxel-wise differences were reflected in whole ROI volume differences, and to generate percent difference estimates. We derived hippocampal, caudate and accumbens volume estimates by integrating image intensity values of modulated and warped grey matter images within the hippocampus, caudate and accumbens masks, respectively. We derived ventricle estimates by integrating image intensity values of modulated and warped cerebral spinal fluid images within the ventricle masks. Because the grey matter tissue of the putamen was only partially classified, we derived putamen volume estimates by integrating intensity values of modulated and warped binary image volumes within the putamen mask.
We used the Statistical Package for the Social Sciences (SPSS) to analyze demographic and volumetric data. We tested the distribution of all continuous data for normality with the Shapiro–Wilk test. Age, duration of untreated illness, cerebral spinal fluid and ventricle volumes were not normally distributed. Logarithmic transformation only normalized the distribution of duration of untreated illness. We tested age, cerebral spinal fluid and ventricle volumes nonparametrically with the Mann–Whitney U test. We tested handedness and sex differences with the Fisher exact test and socioeconomic status with the Pearson χ2 test. Because the Ptab group only consisted of 9 patients, the Mann–Whitney U test was used to compare clinical data (PANSS scores and duration of untreated illness) between the 2 subgroups. We identified potential outliers with the Grubb outlier test.40
We used analysis of covariance to compare Ptall and controls for volumetric estimates of intracranial volume, total brain volume, grey matter and white matter. We entered age, sex and intracranial volume as covariates. Intracranial volume was only corrected for age and sex. We used the Mann–Whitney U test to test for group differences in cerebral spinal fluid volume after the effects of age, sex and intracranial volume had been regressed out. We used the latter approach to test differences for in intracranial volume, total brain volume, grey matter, white matter and cerebral spinal fluid volumes between Ptnon-ab and Ptab.
In the voxel-wise analyses of group differences, age, sex and intracranial volume were covariates. We first tested for differences between Ptall and controls. Subsequently, planned comparisons tested for differences between Ptnon-ab and controls, Ptab and controls, and Ptnon-ab and Ptab. We estimated general linear models nonparametrically using Randomize, version 2.1, part of the FSL library of tools (www.fmrib.ox.ac.uk/fsl/randomise/index.html) with 10 000 permutations. We used small volume correction, applying ROI masks, to test our a priori hypotheses of hippocampal reduction, ventricle enlargement and caudate reduction. A false discovery rate threshold of 0.05 was used to correct for multiple comparisons. The clinical data (positive, negative and total PANSS scores, and duration of untreated illness) were used as covariates in separate analyses.
We analyzed group differences in total hippocampal, ventricles, caudate, accumbens and putamen volume estimates with SPSS using a repeated-measures analysis of variance with group (Ptall and control or Ptnon-ab and control) as the between-subjects factors, and hemisphere (left and right) as the within-subjects variable. Age, sex and intracranial volume were covariates. We tested for volumetric differences of ROIs between Ptab and controls and Ptnon-ab and Ptab by use of the Mann–Whitney U test, after regressing out age, sex and intracranial volume effects. We calculated the percentage ROI volume differences between groups using corrected volumes adjusted for age, sex and intracranial volume.
All tests were 2-tailed, and the significance level was set to p < 0.05.
Of the 43 patients with first-episode schizophrenia, we excluded 5 patients from further analyses (3 whose diagnoses were adjusted to schizotypal personality disorder and 2 with artifacts on their MRI scans). Of the 38 patients included, 9 fulfilled the DSM-IV criteria for lifetime substance abuse. Three of the 9 patients had no history of abuse for the past year, and 5 patients had no abuse for the past month. Diagnoses were based on excessive intake of alcohol (n = 3), cannabis (n = 2), alcohol and cannabis (n = 3) and central stimulants (n = 1). One patient had smoked cannabis on a few occasions in the month before the MRI scan. All participants had a negative urine screening result for substance intake. In total, there were 9 patients with any lifetime substance abuse side-diagnosis (Ptab), 29 patients with no lifetime substance abuse diagnosis (Ptnon-ab). Demographic and clinical characteristics are shown in Table 1.
There were no differences between Ptall and control groups for age (Z = −0.51, p = 0.61), sex (Fisher exact test, p > 0.99), handedness (Fisher exact test, p > 0.99) and parental socioeconomic status (χ22 = 1.85, p = 0.40). Likewise, the 2 patient subgroups, Ptab and Ptnon-ab, did not differ in age, sex, handedness or parental socio-economic status (p > 0.21). Compared with Ptnon-ab, the Ptab group had more benzodiazepines prescribed in the investigation period (Fisher exact test, p = 0.02) and a tendency toward higher PANSS positive scores (Z = −1.84, p = 0.07). There was no difference in antidepressant exposure between the 2 patient groups (lifetime exposure: Fisher exact test, p = 0.66; current treatment: Fisher exact test, p = 0.13), and there was no difference between PANSS negative, PANSS total or duration of untreated illness (p > 0.76).
There were no volumetric differences in intracranial volume, total brain volume, grey matter, white matter or cerebral spinal fluid between the Ptall and control groups (p > 0.23) or between the 2 patient subgroups (p > 0.14) (Table 1).
The voxel-wise analyses revealed significant bilateral reductions in hippocampal grey matter in the Ptall group as compared with the control group. This difference was accounted for by Ptab rather than by Ptnon-ab (Fig. 1). Direct comparison of the 2 patient groups showed significantly reduced hippocampal grey matter in the Ptab group.
The results of the volumetric analyses paralleled those of the voxel-wise analysis. The significant main effect of group, Ptall × control (F1,76 = 4.26, p = 0.042), was accounted for by Ptab (Ptab × control; Z = −3.28, p = 0.001), rather than by Ptnon-ab (Ptnon-ab × control; F1,67 = 0.63, p = 0.43). Moreover, those in the Ptab group had significantly smaller hippocampus volumes than those in the Ptnon-ab group (Z = −2.73, p = 0.006). The comparison of Ptall and control did not reveal a hemisphere × group interaction (F1,76 = 2.61, p = 0.11) or hemisphere effect (F1,76 = 0.31, p = 0.58). The group differences represented corrected hippocampal volume reductions of 2.4% between Ptall and control, 6.5% between Ptab and control, and 0.1% between Ptnon-ab and control (Fig. 2).
Neither the voxel-wise nor the volumetric analyses revealed ventricle differences between any of the groups (p > 0.38 in the volumetric analyses). Grubbs test detected one outlier in the control group. Visual inspection of this participant’s MRI scan indicated no pathology or artifact. Exclusion of this outlier did not change the results. The results of exploratory analyses of group differences for the third and the lateral ventricle volumes separately were not significant.
The voxel-wise analyses revealed significantly reduced caudate grey matter bilaterally in the Ptall group compared with in the control group. This difference was primarily because of Ptnon-ab rather than Ptab (Fig. 3) However, the 2 patient groups did not significantly differ from each other.
The analyses of caudate nucleus volume paralleled the voxel-wise results. There was a significant main effect of group, Ptall × control (F1,76 = 8.94, p = 0.004). This group difference was primarily accounted for by Ptnon-ab (Ptnon-ab × control; F1,67 = 6.88, p = 0.01), although there was a tendency toward caudate volume reduction in Ptab (Ptab × control; Z = −1.90, p = 0.06). There was no volumetric difference between the 2 patient groups (Z = −0.17, p = 0.99). There was no hemisphere × group interaction when comparing the Ptall and control groups (F1,76 = 0.62, p = 0.44) or hemisphere effect (F1,76 = 3.24, p = 0.076). The group differences represented corrected caudate volume reductions of 5.1% between Ptall and control, 5.1% between Ptnon-ab and control, and 5.0% between Ptab and control (Fig. 4).
Exploratory voxel-wise analyses of the nucleus accumbens revealed a tendency for reduced grey matter in the Ptall group compared with the control group. This tendency appeared to be driven by Ptnon-ab, rather than by Ptab. However, the 2 patient groups did not significantly differ from each other. Analyses of nucleus accumbens volume showed a significant main effect of group (Ptall × control; F1,76 = 6.82, p = 0.011). This group difference seemed to be accounted for by Ptnon-ab (Ptnon-ab × control; F1,67 = 7.58, p = 0.008), rather than by Ptab (Ptab × control; Z = −0.88, p = 0.38).
Voxel-wise exploratory analyses of the putamen did not reveal any differences between Ptall and control, Ptnon-ab and control or Ptab and control. Exploratory analyses of putamen volume did not reveal any group main effects (Ptall × control: F1,76 = 0.42, p = 0.52; Ptnon-ab × control: F1,67 = 0.044, p = 0.83; Ptab × control: Z = −1.27, p = 0.20). When comparing the hemisphere × group interactions for the Ptall and control groups (F1,67 = 1.25, p = 0.27 and F1,67 = 0.001, p = 0.97 for the accumbens and putamen, respectively). Hemisphere effects (F1,67 = 0.18, p = 0.73 and F1,67 = 0.012, p = 0.33 for the accumbens and putamen, respectively) were absent.
Exploratory voxel-wise whole brain analyses did not reveal any group differences in regional grey matter, white matter or cerebral spinal fluid volumes.
Neither the voxel-wise nor the volumetric analyses revealed significant associations between PANSS scores or duration of untreated illness and hippocampal, ventricle or striatal structure volumes. However, we observed a weak tendency of longer duration of untreated illness to be associated with reduced hippocampal volume in the Ptall group (F1,32 = 2.55, p = 0.13). Associations with other brain regions, as examined with exploratory whole brain voxel-wise analyses, were absent. There were no significant associations between clinical measures and global brain measures (intracranial volume, total brain volume, grey matter, white matter or cerebral spinal fluid).
As we had hypothesized, this study revealed significant hippocampal and caudate volume reductions in a relatively large cohort of antipsychotic-naive first-episode schizophrenia patients compared with matched healthy controls. Ventricular enlargement was absent. No differences in global volumes were found, and no significant associations between tissue volumes and psychopathology or duration of untreated illness were observed.
Our data support the growing body of evidence that indicates that the hippocampal volume is significantly reduced at the onset of schizophrenia.3,4,6 The observed corrected volume reduction of 2.4% in the schizophrenia patients, Ptall, compared with the control participants was somewhat smaller than the 8.2% reduction reported in a recent meta-analysis of data from first-episode schizophrenia patients.4 However, the patients in the present study had never taken antipsychotics, whereas most of the studies included in the meta-analysis included medicated patients.4 It has been suggested that previous exposure to first-generation antipsychotics may not protect against progressive reductions in hippocampal volume reduction.8 Moreover, the meta-analysis also included studies that did not account for substance abuse or dependence. Interestingly, the hippocampal volume reduction in our study appeared most pronounced in patients with a lifetime substance abuse diagnosis (Ptab), accounting for a corrected volume reduction of 6.5% as compared with controls, whereas the reduction among patients without a history of abuse (Ptnon-ab) was only 0.1%. However, this study was not a priori designed to test for potential effects of substance abuse; thus, interpreting the results about the effects of abuse must be done cautiously.
Indeed, the absolute, uncorrected volumes (Table 3) suggest that both patient groups might have reduced hippocampal volumes compared with controls. Our findings of higher PANSS positive scores and more frequent prescriptions of benzodiazepines among the Ptab patients suggests that stress could also have influenced hippocampal volumes.7 To our knowledge, no studies have associated benzodiazepine use with volumetric brain changes in schizophrenia. Finally, genetic variants prevalent in the normal population may contribute to morphologic variations in schizophrenia;41 hence, the differences between the small patient subgroups observed in this study could reflect random genetic profiles rather than past substance abuse.
Our finding of reduced caudate nucleus volume agrees with studies reporting absolute15 or significant9,25,26,29,42,43 caudate reductions in antipsychotic-naive schizophrenia patients. Thus, our findings add to the evidence that the caudate nucleus is a key structure in the pathophysiology of schizophrenia.44 The pathway to the observed volumetric reductions in the antipsychotic-naive state is not clear but may be attributable to decreased metabolic rates in the basal ganglia.45 The magnitude of the caudate volume reductions in this study of about 5% are in line with the average caudate reduction in the study by Glenthoj and colleagues.15 In the present study, caudate volume reductions were significant in the Ptnon-ab group and were also apparent in Ptab, suggesting only a modest, if any, effect of abuse on caudate volumes.
Ventricular enlargement has consistently been observed in first-episode schizophrenia3,4 and has also been reported in antipsychotic-naive patients.39,43 Nevertheless, ventricular enlargement was absent in the present cohort, and, as such, our data suggest that ventricular enlargement may not occur until a later stage of the disease13 or may be related to anti-psychotic medication use.14
In our voxel-wise analyses, the nucleus accumbens volume appeared reduced at a trend-level in the Ptall group as compared with the control group. However, in the volumetric analyses, this reduction was significant. Volume reductions in the nucleus accumbens have previously been found in antipsychotic-naive schizophrenia patients,15,18,19 but, to the best of our knowledge, significant accumbens volume reductions have not been reported. Still, limited conclusions can be drawn from our finding because it emerged from exploratory analyses. Moreover, the accumbens reductions were only partially supported by the voxel-wise analysis. In agreement with most15,18,25,27,46 but not all19,26 reports on the putamen, we also observed decreased absolute, uncorrected putamen volumes in patients; however, this was not significant when corrected for age, sex and intracranial volume. Only 1 study reported putamen volumes to be significantly reduced,18 rendering the issue of structural changes in the putamen in antipsychotic-naive schizophrenia patients unresolved.
We observed no significant differences in corrected global brain volumes (intracranial volume, total brain volume, grey matter, white matter and cerebral spinal fluid) in the patients as compared with the controls. The absence of total brain volume reduction in the patients is in contrast to the findings by Steen and colleagues4 and could reflect that some of the patients included in the meta-analysis had been exposed to first-generation antipsychotics, which may attenuate global grey matter loss.47
It is unclear whether reductions in the hippocampus and caudate nucleus in antipsychotic-naive schizophrenia are associated with psychopathology and illness duration. Although not significant, our observation of a weak association between longer duration of untreated illness and reduced hippocampal volume is in line with those of Matsumoto and colleagues,11 suggesting that hippocampal changes may occur during the transition to psychosis.48 Associations between structural abnormalities and clinical variables presumably reflect underlying pathophysiologic disturbances in neurotransmission, metabolism and genetic variance, and different mechanisms may underlie positive and negative symptoms.49 Studies investigating the relations between volumetric measures and clinical variables have yielded inconsistent results likely attributable to differences in clinical samples, image acquisition and processing and anatomic delineation protocols.
We found no altered asymmetry between patients and controls, hence we have not replicated our previous finding of altered asymmetry of the caudate nucleus in an independent group of 19 antipsychotic-naive patients.15
The present study is limited with respect to elucidating the possible role of drug abuse on volumetric measures because it was not a priori designed to test for differences between patients with or without a lifetime substance abuse diagnosis. Moreover, even though we have assessed all DSM-IV diagnoses with a validated instrument (SCAN 2.1), we cannot exclude the possibility that some of the patients in the Ptnon-ab group may have had a past period of excessive drug use, which we were not informed of. Additionally, because of the small number of patients with abuse and the various types of abuse, we cannot make causal inferences about the impact of lifetime substance abuse on brain structure in first-episode schizophrenia patients. Nevertheless, comorbid substance abuse has previously been associated with morphologic changes, including in the hippocampus.22,23
Another limitation was the unequal size of the patient subgroups. This was addressed by the use of nonparametric statistical analyses, which makes no assumptions about the distribution of the data. Notably, the volumetric analyses confirmed the differences detected by the voxel-wise analyses, hereby ruling out the possibility that the voxel-wise results were based on random local minima.
Our study indicates that reduced hippocampus and caudate nucleus volumes may constitute morphologic traits in first-episode antipsychotic-naive schizophrenia patients. However, the clinical implications of these findings are still unclear. Moreover, a history of substance abuse may accentuate hippocampal volume reductions. Magnetic resonance imaging studies explicitly addressing the potential effects of any lifetime substance abuse in antipsychotic-naive first-episode schizophrenia patients are warranted.
This study was sponsored by The Danish Medical Research Council, The Copenhagen Hospital Cooperation, The Lundbeck Foundation, Gerda and Aage Haensch’s Foundation, Slagtermester Max Worzner og hustru Inger Worzner’s Foundation and The Danish Psychiatric Association. An unrestricted grant was received from AstraZeneca A/S, Denmark.
Trial Registration Clinicaltrials.gov NCT00207064, http://clinicaltrials.gov/ct2/show/NCT00207064
Competing interests: None declared.
Contributors: Drs. Ebdrup, Glenthøj, Paulson and Baaré designed the study and wrote the article. Drs. Ebdrup, Glenthøj, Rasmussen, Aggernaes and Baaré acquired the data, which Drs. Ebdrup, Glenthøj, Rasmussen, Langkilde, Lublin, Skimminge and Baaré analyzed. All authors reviewed the article and approved its publication.