BACKGROUND

Having previously demonstrated grossly apparent parietal sulcal widening in a subgroup of autistic people [Courchesne & al. 1993], we recently found a narrowing of the posterior corpus callosum in autism [Egaas & al. 1995]. This combination of cortical and callosal deficiencies suggests a loss of posterior cortical neurons including callosally projecting pyramidal cells. It does not, however, rule out the possibility that the callosal thinning and the cortical sulcal widening may be distinct phenomena. In particular, the sulcal widening might arise from loss or agenesis of tissue other than pyramidal cells, and the callosal thinning might arise from loss or partial agenesis of the myelin surrounding callosal axons, rather than the axons themselves. To test this latter possibility, we compared NMR signal intensity in the corpora callosa of autism patients and normal controls.

Myelination of the corpus callosum is microscopically visible at about the time of birth [Gilles & al. 1983]. It becomes grossly evident at about the fourth postnatal month and proceeds anteriorly from the splenium toward the genu [Yakovlev & Lecours 1967]. The development of callosal myelin is associated with an increase in T1 relaxation time [Valk & van der Knapp 1989] as well as a jump in cross-sectional area. The increase in intensity is so pronounced, in fact, that the corpus callosum in young infants appears darker in NMR scans than the cerebral white matter--the reverse of the adult pattern. Conversely, demyelination due to advanced multiple sclerosis is associated with decreases in NMR signal and in cross-sectional area [Laissy & al. 1993]. Non-quantitative assessments of callosal myelination can be misled by the adult appearance of the callosum, which is attained by about eight months of age [Valk & van der Knapp 1989].

Yakovlev and Lecours [1967] on the basis of myelin staining at autopsy reported continuing callosal myelination through the end of the first decade and suggested that slow myelination proceeds even after this stage. Careful quantitative study of the corpus callosum at a wide range of ages has confirmed that callosal myelination continues into the third decade of life. Byne & al. [1988], reporting on a sample of adults, found no significant effect of age on cross-sectional area, but did find a significant difference between groups on partitioning the population into those older than forty years and those younger than forty years. Clarke & al. [1989], again measuring cross-sectional area, found well-defined age progressions in infancy and from adolescence to adulthood. Allen & al. [1991] found a significant age-related increase in cross-sectional area of all subregions of the corpus callosum in children under fifteen years of age. Cowell & al. [1992] also found a significant effect of age on area, and found that area was maximal at about twenty years of age in males, and about fifty in females. Pujol & al. [1993], on the basis of longitudinal examination of a cross-sectional sample of ages, confirmed that significant growth in cross-sectional area persists into the third decade. Our own pilot study of 4 autism patients and 4 normal controls agrees with this conclusion.

Thus there has been extensive use of NMR imaging to elucidate the morphometric progression of the corpus callosum with age, but intensity has rarely been utilised except as an aid to the segmentation of gross morphology. This is not wholly unexpected, because the use of quantitative intensity data presents several pitfalls. First, though morphometric data can be compared across different scan protocols, intensity measures are sensitive to the degree of excitation and decay that occurs before sampling. Second, even when the protocol is fixed, day-to-day variations in tuning of the instrument seem to add a great amount of noise to the measurements. Third, intensity differences in the structures of interest may be very small in comparison to the absolute intensities.

These analytical challenges are considerable, but not insurmountable. With careful consideration, a protocol general enough for long-term use in a large study can be developed. Though variations in signal intensity can be as small as one or two percent, the success of functional NMR imaging establishes the feasibility of extracting information from such small variations. Intensity measurements are a valuable complement to morphometry, because the two types of measure do not necessarily vary in the same way.

Methods