Diffusion tensor brain MR imaging in X-linked cerebral adrenoleukodystrophy

Methods.

Between October 11, 1999 and August 14, 2000, 11 boys (aged 7 to 19 years) with biochemically proven X-linked ALD were evaluated with DT MRI. Two of them were examined twice. All of the patients were participants in a special dietary trial (with Lorenzo’s oil). Ten of these patients had adrenal insufficiency and received steroid replacement therapy. On neurologic and neuropsychological testing, four patients had motor, cognitive, behavioral, visual, and auditory impairments. Institutional review board approval and informed consent were obtained for our study.

MRI was performed on 1.5T clinical MR system (ACS-NT; Philips Medical Systems, Best, the Netherlands). Whole brain DT MRI was obtained using a gradient- and spin-echo read-out with cardiac gating. Acquisition parameters were 3,200 to 5,250 (depending on subject’s heart rate)/119 [repetition time (TR) ms/effective echo time (TE) ms], five gradient echoes per radiofrequency (RF) echo, four RF refocusing periods per segment and four segments, 23 cm field of view, 5 mm slice with 1 mm gap, and 128 × 80 scan matrix. At each level, DT MR images with diffusion sensitization of b-value = 599 s/mm² in six different noncolinear directions and T2-weighted (b0) image without diffusion sensitization were obtained.

DT images and b0 images were transferred to a UNIX workstation. Numerical calculations were made using homemade routines developed in IDL (Interactive Data Language; Research Systems, Boulder, CO). The six independent variables (D_xx, D_yy, D_zz, D_xy, D_yz, D_zx) in the diffusion tensor were calculated from the DT images and b0 images.4 From the diffusion tensor data, voxel-by-voxel brain maps (256 × 256 matrix) of ADC_i ([D_xx+D_yy+D_zz]/3) and FA6 were generated ( figure 1, d and e; figure 2A, e and f).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Fifteen-year-old symptomatic boy with X-linked adrenoleukodystrophy. T2-weighted (3,429/119/0; repetition time ms/effective echo time ms/b-value s/mm²) image (a) and D − W_i (3,429/19/599) images (b and c) obtained at the level of the centrum semiovale demonstrate confluent and symmetric abnormalities in the deep white matter of both parietal lobes. Corresponding apparent diffusion coefficient (ADC_i) (d) and fractional anisotropy (FA) (e) maps were calculated. Area A (A in c), in the periphery of the white matter abnormality, showed slight increase in signal intensity on T2-weighted image and prominent increase in signal intensity on D − W_i. Area B (B in c), in the center of the white matter abnormality, showed prominent increase in signal intensity on both T2-weighted and D − W_i. Area C (C in c) showed prominent increase in signal intensity on T2-weighted image and slight increase in signal intensity on D − W_i. On the FA map, there was a mild decrease signal intensity (i.e., anisotropy) from normal-appearing white matter (NAWM) to area A, and a marked decrease from NAWM to areas B and C.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Seven-year-old boy with gait disturbance. (A) Initial evaluation. D − W_i (4,000/119/599) images (a to c) with different segmented areas, and spin-echo T1-weighted image (517/14) with gadolinium enhancement (d) obtained at the level of the centrum semiovale. Corresponding apparent diffusion coefficient (ADC_i) (e) and fractional anisotropy (FA) (f) maps were calculated. Areas A (A in a), B (B in b), and C (C in c) located from periphery to core, and from the anterior to posterior part within the affected white matter. Curvilinear gadolinium enhancement on the T1-weighted image (arrowheads in d) located along the inner edge of area A. (B) Six-month follow up study performed because of worsening gait disturbance and new onset left hemiparesis. Original ADC_i and FA maps (a and b) and 6-month follow-up ADC_i and FA maps (c and d) obtained at similar levels (lateral ventricles) were compared. On ADC_i maps (a and c) the affected white matter areas of both cerebral hemispheres have increased in size over the 6-month interval (white arrowheads). On FA maps, the areas with decrease signal intensity (loss of anisotropy) in both parietal lobes have enlarged to involve the subcortical white matter (★) over the 6-month interval (b and d).

All the generated maps were reviewed by an author (R.I.), and slice locations with maps affected by artifacts were eliminated. According to apparent difference in signal intensity (SI) pattern on T2-weighted and isotropic diffusion-weighted [D − W_i = (D − W_xx × D − W_yy × D − W_zz)^1/3] images (see figure 1, a and b), white matter was classified by the author into the following four areas: normal-appearing white matter (NAWM) on both T2-weighted and D − W_i images; area A, slight increase in SI on T2-weighted and prominent increase on D − W_i images; area B, prominent increase in SI on both images; area C, prominent increase in SI on T2-weighted and slight increase on D − W_i image (see figure 1c). To evaluate the age of the white matter lesion, this classification was based on the features of acute and chronic plaques on T2-weighted and D − W_i images in MS.8 The areas were segmented using threshold values which were adjusted by SI profiles and automatic seeding routine written in IDL. The classification and segmentation were then reviewed by another author (F.S.E.), and a consensus was achieved where there was a difference in opinion. Furthermore, in order to evaluate the achievement of appropriate segmentation, SI ratios of the segmented areas within affected white matter to NAWM for each image were calculated and compared using Student’s t-test. The distribution and arrangement of the different areas in the lesion were documented by the two authors in consensus.

ADC_i and FA values of the pixels within the segmented areas on the corresponding ADC_i and FA maps were obtained. The average and SD of the measured ADC_i and FA of the areas were calculated and compared using Student’s t-test.

In one patient, T1-weighted (TR = 517 ms, TE = 14 ms, two averages) spin-echo images matched to scan levels of DT images were obtained before and after gadolinium enhancement and reviewed. In two patients who were examined twice, serial changes of the zonal distribution in affected white matter were evaluated.

Results.

Seventy-five slice locations from 13 examinations were obtained. Six patients (eight examinations) had visually detectable abnormality in the white matter on the T2-weighted and D − W_i images. Three patients had only area B, and three patients (five examinations) had areas A, B, and C. Fifteen slice locations showed coexistence of areas A, B, and C. Coexistence of areas A and B were observed in 14 slice locations, and of areas B and C in one slice location. Nine slice locations had only area B, and the remaining 36 slice locations had only NAWM. Zonal gradation of areas A, B, and C from periphery to core in the affected white matter was observed on 12 of 15 slice locations. Area A was located in the periphery of area B on 12 of 14 slice locations. Area B was located in the periphery of area C on 1 of 1 slice location.

SI ratios for the four areas are reported in the table. There were differences in SI ratios of T2-weighted images for three areas within the affected white matter, and in SI ratio of D − W_i images between areas A and C, and between areas B and C (p < 0.01). There was no difference between areas A and B in SI ratio of D − W_i images (p = 0.13).

A total of 79,487 pixels, 5,630 pixels, 9,791 pixels, and 1,969 pixels were used to calculate average ADC_i and FA values from NAWM, area A, area B, and area C. The averages and SD of ADC_i and FA values for the four areas are reported in the table. There were differences in average FA and ADC_i values for all four areas (p < 0.01).

Curvilinear enhancement in area A on T1-weighted image after contrast injection and serial changes of areas in one patient were observed and illustrated in figure 2.

Discussion.

MRI techniques are assessed based on their ability to demonstrate differences between zones of potentially reversible and irreversible white matter disease in X-linked ALD. Determination of ADC_i and FA values in the affected white matter provides a quantitative measure that is intrinsic to the diseased white matter. These intrinsic measures allow normalization across subjects and MR scanners, and are critical for conducting multicenter trials.

Our results demonstrate that ADC_i and FA values in the affected white matter are significantly different from those in NAWM and reveal a zonal gradation of the values within the affected area from periphery to core, which is similar to histopathological zonal changes described by Schaumburg.2 In figure 2A, the distribution of enhancement, which may reflect perivascular inflammation in Schaumburg’s second zone,9 suggests that the inner edge of area A may correspond to Schaumburg’s second zone. Also, expansion of the area with low FA value (loss of anisotropy) in figure 2B, demonstrates the ability of FA maps to monitor disease progression and supports the notion of further aging in the affected white matter.

The drop in the average FA from NAWM to the peripheral area of the lesion may suggest that the integrity of the myelin sheath and axon contribute to diffusion anisotropy in white matter. On the other hand, the increase in average ADC_i from periphery to core within the affected area suggests that there may be an increase in free water and injury of the structures that restrict water diffusion.7,8⇓

The lack of opportunity to correlate DT MRI with pathologic findings and the absence of a well-established animal model for cerebral X-linked ALD represent our limitation. For confirmation of our results, future correlation of DT and spectroscopic MRI10 may be helpful. Another limitation results from the lack of normative ADC_i and FA values from age-matched controls. This limitation prevents us from assessing the ability of ADC_i and FA maps to demonstrate early derangement in NAWM.