Laterality Effects and Alcoholism
Analysis of variance tested group-by-hemisphere interactions in each bilateral fiber bundle. Significant group effects and hemisphere effects were found for 8 of 10 bundles ( and Supplement 1), but in no case was the group-by-hemisphere interaction significant with family-wise Bonferroni correction for significance. Thus, we found no evidence that alcoholism significantly altered any normal left-right asymmetry for a given bundle. Subsequent analyses combined values for FA or diffusivity from left and right hemispheres and were based on mean values.
Fiber Bundle Integrity and Alcoholism
To identify the effects of diagnosis and sex, DTI metrics from the fornix and each of the 10 bilateral fiber bundles were submitted to group-by-sex ANOVAs, the results of which are presented in Supplement 1.
The mean ± SE for FA values and visual depiction of the fiber bundles are presented in , , and and for ADC, λL, and λT in and . Significant group differences were identified in 6 of the 11 bundles: fornix, internal and external capsules, frontal forceps, superior cingulate bundle, and superior longitudinal fasciculus. Specifically, FA was low in the frontal forceps and superior cingulate bundle, and diffusivity measures were higher in the alcoholic than control groups in all six bundles. Trends were present in occipital forceps and inferior cingulate bundle FA.
Figure 2 Mean ± SE FA of the control men and women and alcoholic men and women for fornix, internal and external capsules, and frontal and occipital forceps fiber bundles, depicted above In axial and sagittal views. ANOVA group effects: †p ≤ (more ...)
Figure 3 Mean ± SE FA of the control men and women and alcoholic men and women for superior and inferior longitudinal fasciculus and superior and inferior cingulate fiber bundles, depicted above In axial and sagittal views. ANOVA group effects: †p (more ...)
Figure 4 Mean ± SE FA of the control men and women and alcoholic men and women for pontocerebellar tract and cerebellar hemisphere fiber bundle, depicted above in axial and sagittal views. Af, alcoholic female; Am, alcoholic male; Cf, control female; Cm, (more ...)
Figure 5 Mean ± SE ADC, λL, and λT of the control men and women and alcoholic men and women for the fornix. ANOVA group effects: †p ≤ 01, *p ≤.005. ADC, apparent diffusion coefficient; ANOVA, analysis of variance; (more ...)
Figure 6 Mean ± SE ADC, λL, and λT of the control men and women and alcoholic men and women for the 10 bilateral fiber bundles. ANOVA group effects: *p ≤.005, **p ≤.001, ***p ≤.0001. ADC, apparent diffusion coefficient; (more ...)
Although across all subjects the sex effect was significant for FA or diffusivity in six fiber bundles, only trends were evident in group-by-sex interactions and then only for two bundles: internal capsule λL and inferior longitudinal fasciculus ADC and λL. There, alcoholic men had disproportionately greater diffusivity than the other groups.
Other Contributors to White Matter Fiber Degradation
Regression analyses tested the hypotheses that among the combined group of alcoholic men and women, lower FA and higher diffusivity would correlate with older age, more lifetime alcohol consumption, a shorter period of sobriety, or lower than normal body mass (indicative of poor nutrition). Family-wise Bonferroni correction required p = .01 in the predicted direction to support these directional hypotheses.
Greater lifetime alcohol consumption, in the alcoholic men but not women, correlated significantly or at trend levels with lower FA in the internal capsule (r = −.30, p = .0193), frontal forceps (r = −.36, p = .005), occipital forceps (r = −.46, p = .0002), and superior longitudinal fasciculus (r = −.42, p = .001) and with higher ADC in the internal capsule (r = .36, p = .0057), external capsule (r = .29, p = .0275), frontal forceps (r = .29, p = .0275), and superior longitudinal fasciculus (r = .36, p = .004). These correlations were confirmed with nonparametric Spearman rank tests.
Given that the alcoholic women had been sober for about twice as long as the alcoholic men and that some men had drunk considerably more than any woman, we derived a subgroup of 40 men and 25 women who were matched in length of sobriety (91 days for the men and 103 days for the women), age (45 years for men, 46 years for women), and lifetime consumption of alcohol (557 kg for men, 517 kg for women. We tested the difference between men and women in regional FA and ADC in each fiber bundle. These t tests revealed a significant disadvantage of alcoholic women over men in FA of the internal capsule (p = .0013), inferior cingulate (p = .0076), superior longitudinal fasciculus (p = .002), cerebellar hemisphere (p = .0032), and pontocerebellar (p = .0078) bundles; women also had higher ADC than men in the two cerebellar systems (hemispheres p = .0005; pontocerebellar p = .0203). By contrast, alcoholic women had lower ADC than alcoholic men in the fornix (p = .0158) and inferior longitudinal fasciculus (p = .0141).
Relative to alcoholics who had never smoked (n = 23), current or past smokers (n = 63) had lower FA in the pontocerebellar [t(84) = 2.16, p = .0336] and cerebellar [t(84) = 2.998, p = .0036] fiber bundles and higher ADC in the occipital forceps [t(84) = 2.316, p = .023] and inferior cingulate bundle [t(84) = 2.482, p = .0151]. Alcoholic men and women were combined in this analysis because the few alcoholic women who had never smoked (n = 4) represented too small a group for smoking-by-sex analysis. Curiously, alcoholic smokers had lower ADC in fornix than nonsmokers [t(84) = 3.231, p = .0018].
Body mass index, which was lower in alcoholic women than alcoholic men, was negatively correlated with ADC in the external capsule (r = −.44, p = .0023), frontal forceps (r = −.39, p = .0413), inferior longitudinal fasciculus (r = −.522, p = .0044), and cerebellar hemispheres (r = −.49, p = .0086) in the women but showed no such correlation in the men.
Half of the alcohol sample had a prior history of substance abuse, most commonly cocaine. Women were as likely to be substance abusers as men. Substance-abusing alcoholics had lower FA in the pontocerebellar [t
= .005] and cerebellar hemisphere [t
(85) = 2.59, p
= .011] bundles, greater ADC in the occipital forceps [t
(85) = 2.45, p
= .016], and higher λL in the superior longitudinal fasciculus [t
(85) = 2.49, p
= .015] than nonsubstance-abusing alcoholics.
Cognitive and Motor Performance Associations with DTI Metrics
Group-by-sex ANOVA indicated that although in some cases, women performed better than men (digit symbol output and time to completion) and in other cases men performed better than women (fine finger movement output), in no case was the group-by-sex interaction significant. Significant group differences indicated that compared with control subjects, alcoholics showed deficits in balancing on one foot with eyes open [F(1,161) = 33.934, p = .0001] and closed [F(1,161) = 51.956, p = .0001], fine finger movement output for unimanual [F(1,161) = 5.582, p = .0193] and bimanual conditions [F(1,161) = 5.281, p = .0228], digit symbol traditional output score [F(1,161) = 16.025, p = .0001], and time for grid completion [F(1,161) = 7.779, p = .0059] but not on the unannounced recall test of the symbol [F(1.16) = 1.25, p = .2653].
Given the lack of significant group-by-sex interactions in performance, correlations were conducted on the combined group of alcoholic men and alcoholic women. Among alcoholics, greater lifetime alcohol consumption was associated with lower scores on the traditional digit symbol test (r = −.33, p = .0022; Rho = −.37, p= .0008), time for grid completion (r = −.17, p = .1196; Rho = −.30, p = .0062), and balancing with eyes open (r = −.22, p = .0374; Rho = −.21, p = .0538).
Correlations (Supplement 1) in the predicted direction meeting family-wise Bonferroni correction (p ≤ .01) were present between digit symbol score and FA in the frontal and occipital forceps and ADC in the frontal forceps and internal and external capsule. Time to complete the digit symbol grid correlated with FA in the inferior cingulate bundle and ADC in the fornix, internal and external capsules, and frontal forceps. Fine finger movement output correlated with inferior cingulate FA, and balancing with eyes closed correlated with fornix ADC.