Normal-appearing brain tissue MTR histograms in clinically isolated syndromes suggestive of MS

Methods.

Forty patients with a recent onset CIS, 13 patients with a remote episode of CIS that occurred 14 years previously and did not have MS, and 66 sex- and age-matched control subjects with no previous history of neurologic disease were analyzed. Local ethical committee approval and written informed consent from all subjects were obtained.

2d-SE MTI was performed using a 1.5 Tesla GE system (General Electric Medical Systems, Milwaukee, WI).8 MTR was calculated for each pixel by the formula ([M₀ − M_S]/M₀) × 100 percent units (pu).9 The T2-weighted images were automatically segmented into probability images representing GM, WM, and CSF using SPM99 software (Wellcome Department of Cognitive Neurology, Institute of Neurology, Queen Square, London, UK).10 The total intracranial (TI) volume was defined as GM plus WM plus lesion plus CSF volumes. The brain parenchymal fraction (BPF) was defined as the GM plus WM plus lesion volumes all divided by the TI volume; the gray matter fraction (GMF) was defined as the GM volume divided by the TI volume; and the white matter fraction (WMF) was defined as the WM plus lesion volumes both divided by the TI volume.

A maximum likelihood algorithm, using a probability threshold of 75% certainty was used to extract the MTR values from the GM and WM probability outputs, eliminating partial volume voxels at tissue boundaries. The lesion MTR voxels were also extracted for those patients with lesions. Voxels whose MTR values were less than 7.5 pu were excluded to further minimize partial volume effects. NAWM and NAGM sets were recombined to create a NABT data set. The MTR histogram normalization and analysis have been described previously.9 Statistical analysis was performed using the Mann–Whitney nonparametric test.

Results.

The recent onset CIS cohort had 23 patients with lesions (≥1) on the T2-weighted images; they showed no significant differences in clinical characteristics from the 17 patients with a normal MRI (table 1). NABT mean MTR was decreased by 1.02% compared with matched control subjects (p = 0.001). NABT 25th and 75th percentile MTR were also significantly reduced (table 2). NAWM mean MTR was decreased by 0.64% (37.28 ± 0.36 vs 37.52 ± 0.37 pu, p = 0.009), and a trend was seen for a decrease in the 75th percentile MTR (38.84 ± 0.50 vs 39.10 ± 0.44 pu, p = 0.02) compared with control subjects. There was a decrease in NAGM 50th percentile MTR (32.19 ± 0.52 vs 32.60 ± 0.63 pu, p = 0.002) and trends for a decrease in the mean MTR (31.50 ± 0.51 vs 31.80 ± 0.55 pu, p = 0.01), peak location (33.03 ± 0.29 vs 33.25 ± 0.44 pu, p = 0.01), and 25th percentile (29.41 ± 0.69 vs 29.75 ± 0.78 pu, p = 0.03) compared with control subjects. There were no significant differences in NABT, NAWM, or NAGM MTR histogram features between recent CIS patients with or without T2-weighted lesions. Whole brain atrophy was not detected. However, mean WMF was decreased compared with control subjects (0.352 ± 0.021 vs 0.363 ± 0.018, p = 0.02).

The remote onset CIS cohort had five patients with T2-weighted lesions (see table 1). There was a decrease in NABT mean MTR of 1.45% compared with control subjects (p = 0.009) (table 3). NAWM mean MTR (37.30 ± 0.38 vs 37.59 ± 0.35 pu, p = 0.025) and mean WMF (0.350 ± 0.018 vs 0.367 ± 0.018, p = 0.02) was decreased compared with control subjects. There were no significant differences for NAGM MTR, BPF, or GMF.

Discussion.

Patients with CIS have MTR evidence of subtle damage to NABT within 4 months of the clinical demyelinating episode. It is seen in patients regardless of the presence of T2-weighted lesions and can still be detected 14 years later, in those patients who did not develop clinically definite MS according to the Poser criteria (i.e., further relapses separated in time and space).3 Furthermore, using segmented images, some subtle MTR abnormalities can be detected in both the NAWM and NAGM in the same cohort.

Histogram analysis appears more sensitive than region of interest analysis probably due to more extensive survey of normal-appearing tissues and the larger sample size.5 An earlier MTR histogram study in 11 patients with CIS6 only looked at 2 histogram features—peak height and peak location—and did not find any significant decrease compared with a control group. Histogram peak location but not peak height was decreased in our study of 40 patients with CIS.

The subtle decrease in NABT mean MTR in recent CIS (1% decrease) is similar to that seen in the remote CIS cohort who did not develop MS after 14 years (1.4% decrease). Patients with CIS without T2-weighted lesions are at a lower risk of developing clinically definite MS than patients with CIS with lesions,4 and these patients had a similar decrease as seen in those with a high risk (T2-weighted lesions present). These data suggest that patients with CIS have similar NABT MTR abnormality whether they develop clinically definite MS (i.e., further relapses) or remain static over time. The NABT MTR abnormalities may indicate susceptibility to experiencing a demyelinating event, but not for disease progression. Alternatively, the NABT MTR changes in CIS may represent the sequelae of a single or brief episode of demyelination and inflammation. Thus, both high and low risk patients with CIS would have similar NABT MTR at presentation. Over time, those patients with CIS that have more clinical disease activity may accumulate an increasing severity of NABT MTR abnormality, whereas those patients that remain clinically and MRI static over 14 years lack a progression of NABT MTR abnormality.

The subtle abnormalities in the NABT MTR histograms in CIS could be caused by a combination of tissue abnormalities. First, both WM and GM pathology, as suggested by the subtle abnormalities of some NAWM and NAGM MTR histogram features, could individually or in combination cause a decrease in the NABT MTR values. Second, subtle WM atrophy in the absence of corresponding GM atrophy could of itself contribute to a decrease in NABT MTR by reducing the ratio of WM to GM (because GM MTR is lower). The decrease in WMF but not GMF supports this as a contributory mechanism. Third, partial tissue volumes between brain and CSF in the presence of global atrophy could influence NABT MTR values. However, BPF was not significantly reduced, and the threshold algorithm eliminated partial volume voxels at tissue boundaries.

The possible pathologic basis for MTR abnormalities in the NAWM include edema, astrocytic proliferation, microglial activity, perivascular inflammation, shadow plaques, and Wallerian degeneration. MTR changes in NABT probably become more marked with disease progression and this may reflect a changing pathologic basis—for example, with increasing axonal loss in MS lesions over time, more extensive Wallerian degeneration would occur in the NABT. Small percentage changes in NABT are thus likely to be pathologically nonspecific.