Temporal Lobe Sulco-Gyral Pattern Anomalies in Schizophrenia: An in vivo MR Three-Dimensional Surface Rendering Study

Transcription

1 Temporal Lobe Sulco-Gyral Pattern Anomalies in Schizophrenia: An in vivo MR Three-Dimensional Surface Rendering Study Ron Kikinis*, Martha E. Shenton, Guido Gerig, Hiroto Hokama, Jennifer Haimson, Brian F. O'Donnell, Cynthia G. Wible, Robert W. McCarley, Ferenc A. Jolesz R. Kikinis, F.A. Jolesz, Department of Radiology, MRI Division, Brigham & Women's Hospital, Harvard Medical School, 75 Francis Street, Boston, MA M.E. Shenton, R.W. McCarley, H. Hokama, J. Haimson, C.G. Wible, B.F. O'Donnell, Department of Psychiatry 116A, Harvard Medical School and Brockton VAMC, 940 Belmont Street, Brockton, MA G. Gerig, Institute for Communication Technology, Image Science Division ETH-Zentrum, CH-8092 Zurich, Switzerland. This work was supported by the Whitaker Foundation (RK), the Stanley Foundation (MES, RK), an NIMH Research Scientist Development Award K01-MH00746 (MES), a Scottish Rite grant (MES), a Medical Research Service of the Department of Veterans Affairs (RWM), an NIMH 40,799 (RWM), and by the Commonwealth of Massachusetts Research Center (RWM). *To whom correspondence should be addressed. Key Words: MRI, 3D surface renderings, schizophrenia, sulco-gyral pattern, temporal lobe, neurodevelopment. Summary Neuroanatomical and histological findings from post-mortem brains, as well as in vivo findings from magnetic resonance imaging (MRI) studies, suggest the presence of morphologic temporal lobe abnormalities in schizophrenia. To determine whether or not sulco-gyral pattern abnormalities in the temporal lobe could be detected in vivo, we applied computerized surface rendering techniques to MR data sets in order to make both qualitative and quantitative analyses of three-dimensional reconstructions of the temporal and frontal cortex in 15 schizophrenic patients and 15 normal controls. The qualitative analysis, based on a visual classification of the temporal lobe sulco-gyral pattern by 4 raters blind to diagnosis, showed that in schizophrenics there was a more vertical orientation to the sulci in the left temporal lobe, with an interrupted course of sulci due to gyri coursing across the sulci. Normal controls, in contrast, showed a more horizontal orientation with no interruptions. These findings were supported by the quantitative analysis, where more sulcal lines, representing an interrupted course of sulci, were observed in the temporal lobes (more pronounced on the left) in schizophrenics than in normal controls. These data suggest that some of the abnormalities observed in schizophrenia may have their origin in alterations occurring during the course of neurodevelopment when the sulco-gyral pattern is determined. Over the past decade, new tools for investigating the human brain have dramatically increased our ability to visualize and measure brain tissue. These new tools have been crucial in confirming hypotheses about brain abnormalities in disorders such as schizophrenia where brain abnormalities are often subtle and where, consequently, more precise measurement tools are needed to assess the brain. Today, although there is still no clear understanding of either the etiology or pathology of schizophrenia, recent post-mortem and in vivo magnetic resonance imaging (MRI) studies have made it increasingly clear that structural (morphologic) brain abnormalities may play a key role. More specifically, post-mortem investigations of morphometric abnormalities in the brains of

2 schizophrenics show a relative selectivity for temporal and medial temporal lobe structures [e.g., 2-4, 6, 11-12, 16-17, 21-22, 24]. MRI studies have also found temporal lobe pathology that has been localized to volume reductions in neocortical superior temporal gyrus [1, 30] and to limbic system structures, including the amygdala, hippocampus, and parahippocampal gyrus, especially on the left [e.g., 5, 15, 29, 32]. Whether or not these abnormalities originate from a neurodegenerative process and/or a neurodevelopmental abnormality, remains to be answered. One important study that bears on this issue was conducted by Jacob and Beckmann [20]. These investigators reported abnormalities in the sulcogyral pattern of the left temporal lobe in 42 out of 64 post-mortem brains from patients diagnosed as suffering from schizophrenia; the right temporal lobe was not affected. Post-mortem studies showing sulco-gyral pattern anomalies in schizophrenia are nonetheless limited by tissue and volume changes due to death and/or preservation that may complicate the analysis of the sulco-gyral pattern of the cerebral cortex. In addition, specific testing of clinical symptoms is not possible and diagnostic ascertainment is more uncertain. It would therefore be useful to evaluate the cerebral cortex from in vivo MRI studies. Evaluating the cerebral cortex in vivo would also provide an opportunity to evaluate possible ante-onset determination of risk. Such an in vivo analysis, however, has not been possible until recently due to constraints in both MRI hardware and software. With recent advances, including the ability to make three-dimensional (3D) surface reconstructions of the entire cortex of the brain, these analyses can now be done. We here report a novel application of newly developed computerized surface rendering techniques to MRI data sets. This application provides a unique opportunity to exploit the three-dimensional quality of MRI data more fully, and to visualize the cerebral cortex (temporal and frontal lobe) in schizophrenic patients and normal controls. We report an abnormal sulco-gyral pattern in the temporal lobe (more abnormal on the left) in schizophrenic patients. To our knowledge this study represents the first in vivo study reporting sulco-gyral pattern abnormalities in living schizophrenics from MRI scans. Subjects were 15 right-handed, chronic male schizophrenics who were diagnosed by DSM-III-R criteria [14] and had predominantly positive symptoms (thought disorder, hallucinations, delusions, etc.), although they were not specifically selected for this characteristic. These patients were part of an earlier study where localized volume reductions were reported in neocortical superior temporal gyrus and in limbic system structures, including the amygdala, hippocampus, and parahippocampal gyrus [30]. Mean duration of illness was 16 years for this patient sample. For comparative purposes, the control sample was comprised of 15 age (mean age=37.9 years)-, sex-, and handedness- matched normal controls with no major mental disorder in themselves or in their first degree relatives. The groups did not differ in scores on the Wechsler Adult Intelligence Scale-Revised (WAIS-R) information subscale or in parental socio-economic status [19, 33]. [Further details concerning the subject sample are provided elsewhere- 30]. All subjects were screened for disease factors that could affect brain MRI, including substance abuse and neurological illness, and all signed informed consent before participating in the study. As described in detail in our previous reports [23, 30], the MR imaging protocol was done on a 1.5 T SIGNA GE Scanner. We acquired 3-mm spin echo double echo contiguous axial slices of the whole brain. A segmentation algorithm was then used to establish total brain gray and white matter, and total intracranial contents [8]. None of these global measures was statistically significantly different between the two groups; total left and right temporal lobe volume were also not different between the two groups (see details in [30]). Within the temporal lobe, as reported elsewhere [30], there were volume reductions in left amygdala-hippocampal formation, left superior temporal gyrus, and left (and to a lesser extent, right) parahippocampal gyrus.

3 Surface Rendering. A 3D reconstruction algorithm was used to create surface models of the cerebral cortex in order to depict the topography of the brain surface. These procedures are described in previous reports [8, 9, 26]. The patients were imaged in their individual neutral positions. Accordingly, it was necessary to standardize the positioning of the brain surface in the anterior-posterior (AP), leftright (LR), and superior-inferior (SI) orientations. For this purpose the interhemispheric fissure was aligned parallel to both the AP and LR axes of the rendering environment. In the next step, the brain stem was aligned parallel to the SI axis. After aligning each of the brains in this manner, views from a standard orientation were generated for each subject to display the frontal and temporal lobes. This standard orientation, used for all brains, was based on selecting the same "camera view" which was defined by a fixed view feature of the surface rendering program. Thus once the brains were aligned (see descriptions above), a standard orientation was used to view the temporal lobe for the qualitative analyses, and to view both the frontal and temporal lobes for the quantitative analyses. Following the 3D reconstructions, two analyses were performed on the 30 brains (n=15 controls and n=15 schizophrenics); one based on a qualitative rating by 4 raters and one based on a quantitative analysis of counting the sulci in each frontal and temporal lobe. Qualitative Analysis. For the qualitative ratings only the temporal lobe was evaluated. All measurements were made without knowledge of group membership (i.e., normal or schizophrenic) and all were done using the same standard orientation of the brains (see above). Photographs were made of the left and right hemispheres for all 30 brains. Three raters then drew lines following the sulco-gyral pattern for each of the left and right temporal lobes. This was done independently by the 3 raters without knowledge of each others' ratings. The resulting patterns were then presented to four raters (2 new, 2 who did the original line drawings) who were instructed to categorize the brains into one of two categories: 1) a horizontal orientation of the sulci in the temporal lobe and/or a continuous course of sulci (i.e., not interrupted by gyri coursing across the sulci); or, 2) a vertical orientation to the sulci of the temporal lobe and/or the sulci were interrupted by gyri coursing across them. These categories were selected to match the descriptions of post-mortem findings in schizophrenia by Jakob and Beckmann [20, 21]. They, of course, do not represent true dichotomies but are constructed from what is likely a continuum of patterns. Figure 1 shows a representative view of the sulco-gyral pattern in a normal control (Panel A) and in a schizophrenic patient (Panel B). Note the more characteristic horizontal orientation to the sulci in the normal control (Panel A), and the more vertical orientation in the schizophrenic patient (Panel B), where the lines representing the sulci are interrupted by gyri. Reliability for these ratings was computed using the Kappa statistic [10]. For the left temporal lobe, agreement was k= 0.63 (p<0.001) for all 30 cases, k= 0.52 (p < 0.017) for the schizophrenic group, and k= 0.43 (p < 0.006) for the normal controls. For the right temporal lobe, agreement was k = 0.58 (p<0.001) for all 30 cases, k = (n.s.) for the schizophrenic group, and k = (p<0.001) for the normal controls. Agreement was thus not as strong between raters on right temporal lobe measures for the schizophrenic sample. Table 1 shows a Chi Square analysis of the data for the left temporal lobe (Chi Square analyses for the right temporal lobe were non significant). An average of the 4 raters' classification showed that only 2.5 of the schizophrenic patients were classified as having a horizontal orientation to the sulci whereas 12.5 were classified as having a vertical orientation. Conversely, 11 of the normal controls were classified as having a horizontal orientation to the sulci while 4 were classified as having a vertical orientation. Using the criteria of requiring that three or more of the 4 raters agreed on a vertical orientation resulted in: 3 of the schizophrenics were classified as having a horizontal orientation, 12 were classified as

4 having a vertical orientation, while 12 of the normal controls were classified as having a horizontal orientation and 3 had a vertical orientation (Chi Square=10.8, df=1, p < 0.001). Using the converse criteria of requiring that three of the 4 raters agreed on a horizontal orientation to the sulci showed similar results; 2 schizophrenics were classified as having a horizontal orientation, 13 a vertical orientation, and 10 normals were classified as having a horizontal orientation and 5 a vertical orientation (Chi Square=9.73, df=1, p < 0.001). Quantitative Analyses. The second analysis involved quantitative measures of both the left and right temporal and frontal lobes. These analyses were conducted without knowledge of group, region of interest (temporal/frontal), or hemisphere (left/right). 3D surface renderings of the left and right temporal lobes in a standard orientation (see above) were used as the starting point in the analysis. A newly developed algorithm was then used to identify sulci in the images of the 3D surface renderings of the cerebral cortex. This method, based on geometry limited diffusion [35], automatically characterized the dark sulci in the surface renderings (see Figure 1C & 1D). The method was independent of image intensity and of line (sulcal) thickness, and thus was equally sensitive to sulci of different widths. The digital image was interpreted as a topographical map presenting sulci structures as "valleys" of intensity, changing gradually from high intensity (border of sulcus) to low intensity (bottom of sulci). The algorithm utilized slope information; "creases" (i.e., bottom of the sulci) were defined by points where the gradient direction underwent large changes. A locally adaptive diffusion algorithm smoothed the gradients within slope regions but stopped diffusion at the creases. Successive iterations sharpened smooth "valley" structures and enhanced the lineal structures. The line-extracted image was finally converted into a vector-description by fitting straight vectors to the line pattern. Here, the degree of abstraction was controlled by a parameter expressing the maximum distance of points from the approximated vector, and was the same for all cases. The results for the temporal lobe can be seen in Figure 1C (normal control) and 1D (schizophrenic patient) where the lines of the sulci have been abstracted. To detect line lengths, we carried out a polygonal fit to transform the line raster image into a set of vectors, after first identifying the boundaries of the temporal and frontal lobes. Outside structures were "whited out" using a cursor to change unwanted regions into white background. For the temporal lobe, the sylvian fissure was used to separate the temporal lobe from the frontal lobe and the posterior end of the sylvian fissure was used to draw a straight line to divide the temporal lobe from parietal and occipital lobes. The results of this process can be seen in Figure 2A (normal control) and 2B (schizophrenic patient). For the frontal lobe, the sylvian fissure was used to separate the frontal lobe from the temporal lobe and the central sulcus was used to separate the frontal lobe from the parietal lobe. Separate statistical analyses were carried out for sulci within the boundaries of the temporal and frontal lobes. For the statistical analyses, the total number of lines were assessed after the vectors were oriented in the range of angle +/- 45 degrees to the vertical frame of reference of the image (i.e., the lines seen at the far left/right in Figure 1). The resulting line vectors can be seen in Figure 2C (normal control) and 2D (schizophrenic patient) for the left temporal lobe. The mean number of separate lines were statistically significantly different between groups (p < 0.004), with schizophrenic patients having more lines in the left temporal lobe ( ) than was observed in controls ( ). The right temporal lobe also differentiated the two groups (normal mean= ; schizophrenic mean= ), although the mean difference was less than for the left side (p < 0.029). Because all subjects were dealt with identically, we interpret the greater number of lines in the temporal lobe as suggesting more disorganization resulting from interruptions of the normal horizontal course of sulci by vertical gyri/sulci. No differences were observed between groups for the number of lines in the left or right

5 frontal lobe. Summary. Overall, these data suggest that there is an abnormal sulco-gyral pattern in the temporal lobe in schizophrenics compared to normal controls that is observed on in vivo 3D reconstructions of magnetic resonance images of the brain. There was some suggestion that this abnormality was more pronounced on the left. These data thus confirm and extend the findings of Jakob and Beckmann [20] who reported abnormalities in the sulco-gyral pattern of the left temporal lobe in post-mortem brains of schizophrenics. Our findings are also consistent with a developmental origin in that sulco-gyral development occurs during the third trimester, an observation made as early as 1892 by Cunningham [14]. We also note that differences in cortical development between the two hemispheres may leave the left hemisphere more vulnerable to alterations occurring in the third trimester [7, 13, 18, 25, 34, 36]. Additionally, Rakic [28] has reviewed mechanisms controlling brain and sulco-gyral development, including his own experiments on the developing primate brain which have shown that lesions which affected input to regions also affected sulco-gyral patterns in these target regions. Taken together, these findings point to the importance of factors operating during neural development in altering the sulco-gyral patterns. A cautionary note is, nonetheless, important to add: although our data clearly support the presence of sulco-gyral alterations in the temporal lobe in at least a subgroup of schizophrenic patients, they of course can not pinpoint the responsible mechanism(s). If is of interest, however, that other studies by our group [30] show an intercorrelation of volumetric reductions in densely interconnected temporal lobe regions: posterior superior temporal gyrus and the medial temporal lobe structures of hippocampal/ amygdala complex and parahippocampal gyrus. This intercorrelation suggests the possibility of an alteration in the sulco-gyral pattern via input alterations, as found experimentally by Rakic. We of course do not know which region, if any, represents a "primary" abnormality or whether there is an interplay of multiple abnormalities in these interconnected regions. Our findings also seem to have been anticipated by Southard [31] who in 1915 visually evaluated 25 post-mortem brain specimens of schizophrenics and noted, among other findings, an association between auditory hallucinations and temporal lobe sulco-gyral abnormalities. Unfortunately his work was limited by the methods then available and thus it is only now, more than half a century later, that some of Southard's conclusions have been confirmed. Interestingly, one schizophrenic patient in the present study who was categorized by all 4 raters as showing a horizontal orientation to the sulci (i.e., normal pattern), was the one patient who had no auditory hallucinations and had the highest P300 evoked potential amplitude (higher=more normal) in the left temporal region [27] and also had a thought disorder score that was the second lowest of any subject tested (total score=2.27; a score < 5 is within normal limits). We caution that while our findings seem to be robust, they can not be generalized beyond the sample surveyed here and may not apply to all schizophrenic patients; confirmation of these findings in different subject groups will be important. However, the observed abnormalities in the sulco-gyral pattern of the temporal lobe have important implications for the time of origin of schizophrenic pathology since they point to alterations in the course of neural development. References 1. Barta P.E., Pearlson G.D., Powers R.E., Richards S.S., Tune L.E., Auditory hallucinations and smaller superior temporal gyral volume in schizophrenia, Am J Psychiatry 147 (1990) Benes F., Sorensen I., Bird E., Reduced neuronal size in posterior hippocampus of schizophrenic patients, Schizophr Bull 17 (1991)

6 3. Bogerts B., Zur Neuropathologie der Schizophrenien, Fortschr Neurol Psychiatr 52 (1984) Bogerts B., Meertz E., Schonfeldt-Bausch R., Basal ganglia and limbic system pathology in schizophrenia: a morphometric study of brain volume and shrinkage, Arch Gen Psychiatry 42 (1985) Bogerts B., Ashtari M., Degreef G., Alvir J.M., Bilder R.M., Liberman J.A., Reduced temporal limbic structure volumes on magnetic resonance images in first episode schizophrenia, Psychiatry Res 35 (1990) Brown R., Colter N., Corsellis J.A.N., Crow T.J., Frith C.D., Jagoe R., Johnstone E.C., Marsh L., Postmortem evidence of structural brain changes in schizophrenia: differences in brain weight, temporal horn area, and parahippocampal gyrus compared with affective disorder, Arch Gen Psychiatry 43 (1986) Chi J.G., Dooling E.C., Gilles F.H., Gyral development of the human brain, Ann Neurol 1 (1976) Cline H.E., Lorensen W.E., Kikinis R., Jolesz F.A., Three-dimensional segmentation of MR images of the head using probability and connectivity, J Comput Assist Tomogr 14 (1990) Cline H.E., Lorensen W.E., Souza S.P., Jolesz F.A., Kikinis R., Gerig G., Kennedy T.E., 3D surface rendering: MR images of the brain and its vasculature, J Comput Assist Tomogr 15 (1991) Cohen J., A coefficient of agreement for nominal scales, Educational and Psychological Measurement, 20 (1960) Colter N., Battal S., Crow T.J., Johnstone E.C., Brown R., Burton C., White matter reduction in the parahippocampal gyrus of patients with schizophrenia, Arch Gen Psychiatry 44 (1987) Crow T.J., Ball J., Bloom S.R., Brown R., Bruton C.J., Colter N., Frith C.D., Johnstone E.C., Owens D.G.C., Roberts G.W., Schizophrenia as an anomaly of development of cerebral asymmetry: a post-mortem study and a proposal concerning the genetic basis of the disease, Arch Gen Psychiatry 46 (1989) Cunningham D.J., Surface anatomy of the cerebral hemispheres, Dublin: Published at the Academy House and by Hodges, Figgis, & Co. and Williams & Norgate: London and Edinburgh, Diagnostic and statistical manual of mental disorders. 3rd ed. rev.: DSM-III-R., Washington D.C.: American Psychiatric Association, DeLisi L.E., Dauphinais D.I., Gershon E.S., Perinatal complications and reduced size of brain limbic structures in familial schizophrenia, Schizophr Bull 14 (1988) Falkai P., Bogerts B., Cell loss in the hippocampus of schizophrenics, Eur arch Psychiatry Neurol Sci 236 (1986) Falkai P., Bogerts B., Rozumek M., Limbic pathology in schizophrenia: the entorhinal region -- a morphometric study, Biol Psychiatry 24 (1988) Fontes V., Morfologia do Cortex Cerebral (Desenvolvimento). Lisbon: Instituto Antonio Audelio da Costa Ferreira (1944). 19. Hollingshead A.B., Two factor index of social position, New Haven, CN: Yale University Press, Jakob H., Beckmann H., Prenatal developmental disturbances in the limbic allocortex in schizophrenia, J Neural Transm 65 (1986)

7 21. Jakob H., Beckmann H., Gross and histological criteria for developmental disorders in brains of schizophrenics, J R Soc Med 82 (1989) Jeste D.V., Lohr J.B., Hippocampal pathologic findings in schizophrenia: a morphometric study, Arch Gen Psychiatry 46 (1989) Kikinis R., Shenton M.E., Gerig G., Martin J., Anderson M., Metcalf D., Guttmann C.R.G., McCarley R.W., Lorensen W.E., Cline H., Jolesz F.A., Routine quantitative analysis of brain and cerebrospinal fluid spaces with MR imaging, JMRI 2 (1992) Kovelman J.A., Scheibel A.B., A neurohistological correlate of schizophrenia, Biol Psychiatry 19 (1984) LeMay M., Culebras A., Human brain: morphologic differences in the hemispheres demonstratable by carotid arteriography, N Engl J Med 287 (1972) Lorensen W.E., Cline H.E., Marching cubes algorithm: a high resolution 3D surface reconstruction algorithm, ACM Comput Graphics 21 (1987) McCarley R.W., Shenton M.E., O'Donnell B.F, Faux S.F., Kikinis R., Nestor P.G., Jolesz F.A., Auditory P300 abnormalities and left posterior superior temporal gyrus volume reduction in schizophrenia, Arch Gen Psychiatry 50 (1993) Rakic P., Specification of cerebral cortical areas, Science 241 (1988) Rossi A., Stratta P., D'Albenzio L., Tartaro A., Schiazza G., di Michele V, Bolino F., Casacchia M., Reduced temporal lobe areas in schizophrenia: preliminary evidence from a controlled multiplanar magnetic resonance imaging study, Biol Psychiatry 27 (1990) Shenton M.E., Kikinis R., Jolesz, F.A., Pollak S.D., LeMay M., Wible C.G., Hokama H., Martin J., Metcalf D., Coleman M., McCarley RW., Abnormalities of the left temporal lobe and thought disorder in schizophrenia: a quantitative magnetic resonance imaging study, N Engl J Med 327 (1992) Southard E.E., On topographical distribution of cortex regions and anomalies in dementia praecox, with some account of their functional significance. Am J Insanity 71 (1915) Suddath R.L., Christison G.W., Torrey E.F., Casanova M.F., Weinberger D.R., Anatomical abnormalities in the brains of monozygotic twins discordant for schizophrenia, N Engl J Med 322 (1990) Wechsler D., WAIS-R: manual: Wechsler adult intelligence scale- revised. New York: Harcourt Brace Jovanovich, Weinberger D.R., Luchins D.J., Morihisa J., Wyatt R.J., Asymmetric volumes of the right and left frontal and occipital regions of the human brain, Ann Neurol 11 (1982) Whitaker R.T., Gerig G., Geometry-limited diffusion in the characterization of geometric patches in images, Computer Graphics and Image Processing: Image Understanding 57(1) (1993) Witelson S.F., Kigar D.L., Sylvian fissure morphology and asymmetry in men and women: bilateral differences in relation to handedness in men, J Comp Neurol 323 (1992) Figure Legend

8 Figure 1 In each panel, the normal control is shown on the left and the schizophrenic patient on the right. Panel A and B show the 3D surface rendering of the brain surface from MR data for a normal control (left) and a schizophrenic patient (right). These surface renderings are in the same standard orientation (see description of method in the text). Panel C and D show the results of applying geometry-limited diffusion to the 3D surface rendering of the brain; the resulting lines observed represent the locations of sulci (see text). Figure 2 The same two subjects depicted in Figure 1 are also shown here. In each of panel, the normal control is shown on the left and the schizophrenic patient on the right. Panel A and B show the results of applying

9 geometry limited diffusion to the 3D surface rendering of the brain, and then delineating just the temporal lobe sulci for further evaluation (see text). When the temporal lobe is extracted, only the sulci in this area are seen. The final analysis was based on using the vectors within +45 degrees of the vertical frame of reference of the image (Panels C and D). While the orientation and number of lines are both apparently altered (i.e., greater number of lines and more vertical orientation to the lines in the schizophrenic patients), in this study we analyzed only the number of lines. Note the increase in the number of lines for the schizophrenic patient (Panel D) compared to the control subject (Panel C). Acknowledgments The authors gratefully acknowledge the editorial comments by Drs. Margaret Niznikiewicz and Dean Salisbury, and the technical and administrative support provided by I-han Chou, Diane Doolin, Marianna Jakab, Adam Shostack, Maureen Ainslie, Brian and Eva Chiango, Sue Law, and Marie Fairbanks.