The results of this study demonstrate that prenatal alcohol exposure affects white matter integrity in the corpus callosum, specifically in the isthmus and posterior regions and that these effects are most consistently found in individuals who also exhibit the physical characteristics of exposure. As a result of this and several previous studies, the usefulness of DTI as a method for identification of such neurological alterations is confirmed. In our previous study (Ma, et al. 2005
), we included only very affected, dysmorphic individuals and unexposed controls. In the present study, we increased the number of dysmorphic individuals studied and included some who were less severely affected. In addition, we included a group of young adults who were prenatally alcohol exposed and matched for cognitive level but who showed no external physical effects (ETOH). In including this nondysmorphic group, we sought to re-examine the conclusion of other investigators (Wozniak, et al. 2006
) that nondysmorphic but alcohol-exposed individuals exhibit neurological alterations similar to those in dysmorphic individuals. We anticipated that by matching ability level we would assure a similar level of structural and microstructural alterations in brain if these alterations were not related to external physical characteristics.
The present results confirmed alteration of white matter integrity in the DYSM group, including a significant FA decrease in the isthmus and a trend of FA decrease in the splenium and the genu (F(2,79)
=2.884, p=0.062) (). These results are consistent with our previous study, although in that smaller study only the FA in the genu and splenium were investigated with significant differences found between DYSM and control groups (Ma, et al. 2005
). In contrast, no statistically significant differences (Tukey HSD method, p>0.2) of white matter integrity were seen when comparing the nondysmorphic group with the other two groups, respectively. However, the results (see ) suggest that mean of the DTI measures of the nondysmorphic individuals fall between those of the dysmorphic and the unexposed groups. Further research is necessary to understand the pattern of results in relation to both extent of exposure and characteristics of the exposed.
Examination of patterns in the observed white matter abnormalities, particularly AD and RD, found in alcohol-affected individuals may allow inferences about pathology (Basser 1995
; Basser, et al. 2000
). Both axonal loss and demyelination can cause DTI measures to vary but different etiologies are associated with different patterns of directional diffusivity changes. These possibilities were explored by Song and colleagues in a shiverer mice model characterized by incomplete myelin formation but intact axons (Song, et al. 2002
). In these mice, they observed significantly higher RD but unchanged AD in the shiverer mice versus age-matched controls (Song, et al. 2002
). To further test the hypothesis that AD and RD reflect axon and myelin pathologies, they examined a mouse model of retinal ischemia and observed a significant AD change three days after ischemia with no detectable demyelination. Increased RD was observed five days after ischemia, consistent with histological findings of myelin degeneration at that time (Song, et al. 2003
). The potential of RD as a specific assessment of demyelination and remyelination was further supported in a mouse cuprizone model (Song, et al. 2005
). These results indicate that AD and RD might be more sensitive markers than FA or RA (which are summary parameters) in differentiating myelin loss and axonal injury. In the present study, skeleton-based ROI analysis found that, at the isthmus, AD did not differ between DYSM and control groups while RD was elevated in the DYSM group (). Therefore, the pattern of changes in directional diffusivities (AD unchanged, RD increased) suggests demyelination as the main pathological mechanism for white matter abnormalities in DYSM participants. However, this conjecture is tentative and needs to be confirmed with histological studies. Post-mortem studies of chronic alcoholics reported a degradation of brain white matter microstructure including demyelination (Alling and Bostrom 1980
; Kril, et al. 1997
) but no such studies have been done to ascertain the effects of prenatal exposure.
In DTI studies that examine the corpus callosum, findings have been more frequent in the isthmus and posterior regions. It is of interest to speculate whether there is any indication that associated areas of the cerebral cortex support behavioral functions that are altered as a result of prenatal exposure. The relationship of callosal fibers to the division of the cerebral cortex has been studied in non-human primates (Cipolloni and Pandya 1985
; Pandya, et al. 1971
; Seltzer and Pandya 1983
) and in humans(Abe, et al. 2004
; Huang, et al. 2005
). de Lacoste (de Lacoste, et al. 1985
) found that the isthmus contains fibers coursing from the peri-Sylvian region. Recent diffusion MR studies have successfully parcellated corpus callosum based on its connections to the different cortical lobes (Abe, et al. 2004
; Alexander, et al. 1997
; Huang, et al. 2005
) and the results are similar to de Lacoste’s. These MR tractography-based parcellation studies show that, even with large individual differences, the major callosal fibers coursing through isthmus are connected with the sensory-motor (Alexander, et al. 1997
) and parietal cortex (Abe, et al. 2004
; Huang, et al. 2005
; Mazerolle, et al. 2008
). These cortices have been shown to play important roles in interhemispheric visuo-motor integration (Iacoboni and Zaidel 2004
) and transfer of tactile information (Fabri, et al. 2001
). Archibald et al., employed a volumetric approach to study regional pattern of brain hypoplasia resulting from prenatal exposure to alcohol and concluded that the parietal cortexes are particularly vulnerable to the teratogenic effects of prenatal alcohol exposure (Archibald, et al. 2001
). Sowell’s group have made similar observations (Sowell, et al. 2008b
). In neuropsychological studies of behavior in affected individuals, visual/spatial deficits have been documented frequently although not exclusively (Conry 1990
; Mattson, et al. 1998
). Of note is the study by Sowell et al. in which a correlation between decreased FA in the splenium and compromised visuo-motor ability in the alcohol-exposed individuals was observed (Sowell, et al. 2008a
), suggesting an avenue for further investigation.
A review of the very limited literature on DTI in alcohol-exposed individuals suggests substantial similarities in outcomes. The differences that are observed may result from the research designs (Lebel, et al. 2008
; Ma, et al. 2005
; Sowell, et al. 2008a
). In our earlier study(Ma, et al. 2005
), those in the DYSM group had higher dysmorphia scores and lower IQs and, if physical dysmorphia is a predictor of DTI results, difference in outcome can be expected and can be useful in informing our understanding of these phenomena. In addition, ours are the only studies of young adults. Other published studies were of children and adolescents (Lebel, et al. 2008
; Sowell, et al. 2008a
; Wozniak, et al. 2006
) and there may be different developmental trajectories of brain structures as a function of prenatal exposure. Another methodological difference results from the assumptions associated with the use of TBSS. Although it is designed to resolve the alignment inaccuracies in VBM-style approach and aims for higher sensitivity, objectivity and interpretability, TBSS is applied under a general assumption that alterations in the white matter skeleton are representative of the changes in all white matter tracts (Smith, et al. 2006
). This assumption is acceptable, but could be invalid under certain circumstances, causing false-negatives in the statistical inferences. For example, TBSS used in this study may not be able to detect the differences at “the lateral aspects of the splenium of the corpus callosum” reported by Sowell et al.
(Sowell, et al. 2008a
) as only the maximal local FA values at the splenium would be projected onto the representative skeleton and fed into the following statistics. Last, limited DTI data coverage and ROI sampling regions as well as the relatively stringent statistics (cluster-size multiple comparison correction pFWE<0.05
, t>2.5) could reduce the number of significant findings in this study. Further investigations with whole-brain diffusion data are needed to elucidate the differences in the results reported here and those by the other researchers.
Another potential limitation should be considered. The most significant differences in white matter integrity between the DYSM and control groups were found in the isthmus of the corpus callosum, which is the narrowest part of the corpus callosum in the midsagittal plane. Looking closely at the isthmus in the T1 structural and DTI images revealed that even though only the skeletons of the corpus callosum of the midsagittal plane were investigated in the current ROI analysis, partial volume effects might still play a role at our DTI resolution (1.7×1.7×2 mm3). Due to the thinning of corpus callosum suffered by many DYSM participants, the detected difference at this subregion might be a result of both the pathological alteration of tissue integrity and partial voluming. However, in the genu and splenium, partial volume effects are negligible and the trends of reduced FA observed in the DYSM compared with the control group should be mostly due to underlying changes in tissue microstructure. Finally, while the prenatal identification of this sample and the inclusion of a socioeconomic status (SES)-matched control group answers many of the objections that are usually raised regarding potential confounds in clinically identified samples, in human studies it is impossible to control all of the factors that may impact neurodevelopment and some differences among these groups remain including both prenatal exposure to drugs other than alcohol and a variety of postnatal environmental factors.