Development of Hypointense Lesions on T1-Weighted Spin-Echo Magnetic Resonance Images in Multiple Sclerosis: Relation to Inflammatory Activity

Objective  To evaluate whether degree of inflammatory activity in multiple sclerosis, expressed by frequency of gadolinium enhancement, has prognostic value for development of hypointense lesions on T₁-weighted spin-echo magnetic resonance images, a putative marker of tissue destruction.

Design  Cohort design with long-term follow-up. Thirty-eight patients with multiple sclerosis who in the past had been monitored with monthly gadolinium-enhanced magnetic resonance imaging for a median period of 10 months (range, 6-12 months) were reexamined after a median period of 40.5 months (range, 33-80 months).

Setting  Magnetic Resonance Center for Multiple Sclerosis Research, Amsterdam, the Netherlands, referral center.

Main Outcome Measures  The new enhancing lesion rate (median number of gadolinium-enhancing lesions per monthly scan) during initial monthly follow-up; hypointense T₁ and hyperintense T₂ lesion load at first and last visit.

Results  The number of enhancing lesions on entry scan correlated with the new enhancing lesions rate (r=0.64; P<.001, Spearman rank correlation coefficient). The new enhancing lesion rate correlated with yearly increase in T₁ (r=0.42; P<.01, Spearman rank correlation coefficient) and T₂ (r=0.47; P<.01, Spearman rank correlation coefficient) lesion load. Initial T₁ lesion load correlated more strongly with yearly increase in T₁ lesion load (r=0.68; P<.01, Spearman rank correlation coefficient).

Conclusions  Degree of inflammatory activity only partially predicted increase in T₁ (and T₂) lesion load at long-term follow-up. Initial T₁ lesion load strongly contributed to subsequent increase in hypointense T₁ lesion load, suggesting that there is a subpopulation of patients with multiple sclerosis who are prone to develop destructive lesions.

THE INABILITY to predict the clinical course of multiple sclerosis (MS) in individual patients has made it difficult to decide which patients are suitable for early treatment, since some will not develop disability for many years. Some therapies (interferon beta and copolymer 1) have been shown to suppress both relapse rate and the rate of gadolinium (Gd)–enhancing lesions, although a robust effect on disability is more difficult to establish.^1-4

Recent studies have focused on the relationship between accumulation of hypointense lesions ("black holes") on T₁-weighted spin-echo magnetic resonance (MR) images and development of disability^5,6 in patients with MS. At autopsy, these lesions correlate with axonal loss and matrix destruction,⁷ supporting the hypothesis that hypointense lesions may be used as a putative marker of severe tissue destruction in MS. These autopsy findings are in line with recent biopsy findings: hypointensity on T₁-weighted scans was shown to be affected mainly by 2 factors, the extent of axonal reduction and extracellular edema, capable of expanding the extracellular space.⁸ Further, recent recommendations already include change in T₁ lesion load (in addition to change in T₂ lesion load) as an MR imaging outcome measure in definite (phase 3) clinical trials in patients with established MS.⁹

Factors influencing development of hypointense lesions have been obscure. Truyen et al⁶ hypothesized that hypointense lesions may develop as a result of exhaustion of repair mechanisms (eg, remyelination), associated with the secondary progressive phase of the disease. Lassmann et al¹⁰ suggested that exhaustion of repair mechanisms is the consequence of a high degree of inflammatory activity or of repeated inflammation of the same lesions. Gadolinium enhancement is a marker for breakdown of the blood-brain barrier and therefore for inflammation in new and reactivated chronic lesions^11-13; the presence of enhancement is more frequent during relapses and correlates well with clinical activity.^14-16

Since Gd enhancement and hypointense T₁ lesions can be considered MR imaging equivalents of relapse rate and fixed deficit, we investigated whether degree of inflammatory activity, as depicted by frequency of Gd-enhancing lesions, is related to development of hypointense lesions at long-term follow-up.

Thirty-eight patients (20 women and 18 men) with clinically definite MS¹⁷ who had been monitored between February 1, 1990, and November 30, 1993, with monthly Gd-enhanced MR imaging for at least 6 months were reexamined. Informed consent was obtained after the nature of the procedure(s) had been fully explained. These patients previously had been involved in follow-up studies^5,6 and natural course studies,^18,19 and some of the patients (n=24) had been monitored with monthly MR imaging to investigate treatment efficacy of the monoclonal anti-CD4 cM-T412 antibody.²⁰ Nine of these 24 patients were also involved in a recent follow-up study investigating monthly enhanced MR imaging lesion rate and changes in T₂ lesion volume.²¹

Regarding monthly Gd-enhanced MR imaging examinations, 5 patients had been followed up for 6 months, 1 patient for 8 months, 23 patients for 10 months, and 9 patients for 12 months. At the time of the initial examination, 26 patients had relapsing-remitting MS (RRMS) and 12 patients had secondary progressive MS (SPMS); median age at first visit was 33.0 years (range, 23-52 years), and median disease duration was 3.5 years (range, 0-26 years). Full neurologic history and examination at study entry and exit were performed in all cases, with disability assessed by means of the Kurtzke Expanded Disability Status Scale (EDSS).²² Because of the nature of this study, different neurologists performed these examinations at study entry and exit. Patients using immunosuppressive drugs other than infrequent short courses of intravenous methylprednisolone during relapses were excluded. Clinical and MR imaging examinations were not performed within 1 month after treatment with intravenous corticosteroids. Patients who were treated with the monoclonal anti-CD4 antibody cM-T412 (13 patients were receiving active treatment for 6 consecutive months) were not excluded because of lack of treatment effect on degree of MS activity as measured by monthly Gd-enhanced MR imaging.²⁰

The first series of scans were all performed at 0.6 T (Technicare, Solon, Ohio) and the last scan was performed at 1.5 T (Vision; Siemens, Erlangen, Germany) with the use of standard head coils. During the initial phase, median monthly follow-up was 10 months (range, 6-12 months). Median interval between first and last visit was 40.5 months (range, 33-80 months). A comparable MR imaging protocol was used for the first series and for the last scan, including an equal number of slices and equal dose of Gd on both occasions. Identical slice positioning was achieved by means of unenhanced images in 2 or 3 consecutive planes, correcting for positioning differences according to internal landmarks.²³ After administration of gadopentetate dimeglumine (Magnevist; Schering AG, Berlin, Germany) (0.1 mmol/kg), axial T₂-weighted spin-echo images (0.6 T: repetition time, 2755 milliseconds; echo time, 60 and 120 milliseconds; number of excitations, 2; 1.5 T: 2800, 60 and 120, and 1, respectively) and T₁-weighted images (0.6 T: 450, 28, and 4, respectively; 1.5 T: 550, 15, and 2, respectively) were obtained. Nineteen slices with an in-plane resolution of 1.0×1.3 mm (0.6 T) or 0.98×0.98 (1.5 T) and a slice thickness of 5 mm were obtained with an interslice gap of 1.25 mm.

Monthly postcontrast T₁-weighted images were analyzed for number of new enhancing lesions by 2 readers (L.T. and F.B.) by consensus. The new enhancing lesion rate was calculated per patient and defined as total number of new enhancing lesions on monthly MR imaging divided by number of monthly follow-up scans. The entry and exit scans were analyzed by 1 observer (M.A.A.v.W.) who was blinded to clinical data. Hypointense lesions on postcontrast T₁-weighted images obtained at both first and last visit were analyzed and marked on hard copies. Cutoff levels for hypointense lesions was based on visual analysis and defined as regions with low signal intensity compared with surrounding normal-appearing white matter, corresponding to a hyperintense region on T₂-weighted images. Hyperintense lesions on T₂-weighted images assessed at first and last visit were analyzed and marked on hard copies; they were defined as sharply demarcated regions of high signal intensity (separated from surrounding normal-appearing white matter by a distinct border). Both T₂- and T₁-weighted images of the first and last visit were transferred to a computer workstation (Sparc 5; SUN, Palo Alto, Calif) for quantitative analysis. Hyperintense T₂ and hypointense T₁ lesion loads were quantified by a single observer, blinded to clinical data, by means of a seed-growing technique with home-developed software.⁵ T₁ and T₂ lesion volumes were calculated by adding the area of all lesions multiplied by 0.625 (interslice distance=slice thickness+interslice gap). To evaluate intraobserver variation, scans of 5 patients with varying lesion loads were reassessed. The intraobserver coefficient of variance for this method was 2.8% for T₂-weighted images and 3.0% for T₁-weighted images.

Change in T₁ and T₂ lesion load and change in EDSS during the study period for each patient was adjusted for the length of study. Adjusted annual values were entered into the analysis. The new enhancing lesion rate was adjusted for the duration of the initial monthly follow-up to provide monthly enhancing lesion rates for patients (lesions per month per patient). Since most data were not normally distributed, medians were used to describe the data. Spearman rank correlation coefficient (r) was calculated for correlations. All comparisons underwent 2-sided testing with nonparametric tests, with a significance level of .01 to correct for multiple comparisons. Multiple regression analysis (forward stepwise, F to enter=0.05) was used to estimate the main predictive variables for development of hypointense lesions and for progression of clinical disability.

Significant differences were present between patients with RRMS and SPMS regarding age, disease duration, and EDSS score at entry and exit of the study (Table 1). Regarding the total study group, median change in EDSS score per year was 0.16 (range, −1.3 to +1.1). During the study, an increase in EDSS was present in 21 patients; a decrease, in 6 patients; and no change, in 10 patients. In 1 patient EDSS score was not available for 1 data point. Significant increase in disability (increase in EDSS score, >0.5) occurred in 16 patients (42%) during the study (Figure 1).

Thirty-four patients showed enhancing lesions during initial monthly follow-up; in 4 patients no enhancing lesions were observed. For the total study group, median new enhancing lesion rate was 0.86 (range, 0-9.25). The median number of enhancing lesions on the first scan was 0 (range, 0-9). Median new enhancing lesion rate was 1.1 (range, 0-9.25) for patients with RRMS and 0.67 (range, 0-1.7) for patients with SPMS (Mann-Whitney test, P=.30) (Table 2).

The median initial hypointense T₁ lesion load was 1.5 cm³ (range, 0-22.8 cm³) and was significantly higher in patients with SPMS than in those with RRMS (Mann-Whitney test, P<.001). Hypointense T₁ lesion load increased in 37 of 38 patients, with a median increase of 1.6 cm³ (range, 0-12.9 cm³); final hypointense lesion load was 3.4 cm³. In 1 patient with RRMS, no hypointense lesions were present on the initial scan and no hypointense lesions developed during follow-up. Median T₂ lesion load at entry was 10.8 cm³ (range, 1.3-51.8 cm³) and was significantly higher in patients with SPMS (20.6 cm³) than in patients with RRMS (9.2 cm³) (Mann-Whitney test, P<.001). Final T₂ lesion load was 17.0 cm³ (range, 1.2-51.2 cm³). Median increase in T₂ lesion load was 2.7 cm³ (range, −7.4 to +16.7 cm³) (Figure 2). T₂ lesion load increased in 28 patients (21 with RRMS and 7 with SPMS) and decreased in 9 patients (5 with RRMS and 4 with SPMS). For 1 patient, the initial T₂ lesion load could not be analyzed because of data storage problems. For the whole study group, median change in T₁ lesion load per year was 0.74 cm³ (range, 0-6.7 cm³) and median change in T₂ lesion load per year was 0.93 cm³ (range, −2.6 to +7.8 cm³); no differences were present between the subgroups. For the total study group, median initial T₁/T₂ ratio was 0.13 (range, 0-0.56), and final T₁/T₂ ratio was 0.24 (range, 0-0.57). Both initial and final T₁/T₂ ratios were higher for patients with SPMS than for those with RRMS.

Exclusion of patients who were treated with the monoclonal anti-CD4 antibody cM-T412 (n=13) yielded the following MR imaging characteristics: new enhancing lesion rate of 0.9 (range, 0-9.25); absolute difference in T₂ lesion load, 2.8 cm³ (range, −7.4 to +16.7 cm³); absolute difference in T₁ lesion load, 1.6 cm³ (range, 0-12.9 cm³); median change in T₁ lesion load per year, 0.53 cm³ (range, 0-6.7 cm³); median change in T₂ lesion load per year, 0.90 cm³ (range, −2.6 to +7.8 cm³); initial T₁/T₂ ratio, 0.12 (range, 0-0.56); and T₁/T₂ ratio at exit, 0.22 (range, 0-0.53). In comparison with the MR imaging characteristics of all patients, no significant difference was present for any of these MR imaging variables (Mann-Whitney test, P>.10).

Exclusion of patients treated with anti-CD4 antibody cM-T412 did not change the correlations for baseline and follow-up data significantly. Therefore, correlations referring to all patients are described. Number of enhancing lesions on entry scan correlated with the new enhancing lesion rate at monthly follow-up (r=0.64; P<.001). Similarly, the number of enhancing lesions at entry scan correlated with the number of new enhancing lesions in the first 3 months of the study (r=0.87; P<.01). The number of enhancing lesions on entry scan did not correlate with yearly increase in hypointense T₁ or T₂ lesion load during long-term follow-up. For the whole group, new enhancing lesion rate during monthly follow-up correlated (r=0.42; P<.01) with yearly increase in hypointense T₁ lesion load and correlated more with yearly increase in T₂ lesion load (r=0.47; P<.01) (Figure 3). In patients with SPMS, correlations between new enhancing lesion rate and yearly increase in hypointense lesion load differed (r=0.54; P=.07) from those in patients with RRMS (r=0.48; P=.01). For the total group of patients, new enhancing lesion rate did not correlate with yearly change in T₁/T₂ ratio. The new enhancing lesion rate was higher for patients who showed an increase in T₂ lesion load (median, 1.1; range, 0-9.3) than for patients with a decrease in T₂ lesion load (median, 0.3; range, 0.1-1.9). New enhancing lesion rate was only slightly higher for patients who showed a significant increase in EDSS score (median, 1.4; range, 0-3.7) than for patients with an increase in EDSS score of less than 1 full point during long-term follow-up (median, 1.2; range, 0-9.3). No correlation was present between new enhancing lesion rate and yearly change in EDSS score (Figure 4), although a positive correlation was present for patients with SPMS (r=0.50; P=.09). Baseline T₁ lesion load correlated significantly with subsequent increase in hypointense lesion load (r=0.68; P<.01) (Figure 5). The correlation between baseline T₁ lesion load and new enhancing lesion rate showed a trend for patients with SPMS only (r=0.67; P=.02).

Initial EDSS score correlated with initial hypointense T₁ lesion load (r=0.58; P<.01) and to a lesser extent with initial T₂ lesion load (r=0.49; P<.01). The EDSS score at last visit correlated with hypointense T₁ lesion load at end (r=0.46; P<.01), but not with T₂ lesion load at end (r=0.32; P=.05). Initial T₁/T₂ ratio correlated with initial EDSS score (r=0.34; P=.04), and T₁/T₂ ratio at end of the study correlated even more strongly with EDSS score at last visit (r=0.52; P<.01).

Multiple regression analysis (forward stepwise) was performed to determine the main factors predicting hypointense T₁ lesion load at study exit (dependent variable). Independent variables investigated were age, disease duration, EDSS score at entry, initial hypointense T₁ and T₂ lesion loads, and new enhancing lesion rate. This yielded a model with a multiple R value of 0.95 (R²=0.91). The following parameters were included in the model: initial hypointense T₁ lesion load (β=0.97; P<.001), disease duration (β=−0.14; P=.03), and new enhancing lesion rate (β=0.12; P=.03).

Similarly, we determined the main factors predicting final EDSS score at study exit (dependent variable). Independent variables investigated were age, disease duration, EDSS score at entry, initial hypointense T₁ and T₂ lesion loads, absolute and relative changes in hypointense T₁ and T₂ lesion load, the initial T₁/T₂ ratio, and the new enhancing lesion rate. This yielded a model (multiple R=0.63; R²=0.39) that included the EDSS score at entry (β=0.63; P<.001) as the only variable.

In this study we found correlations between clinical disability and hypointense lesion load at study entry and exit. These correlations were stronger than for T₂ lesion load. Disability correlated with the ratio of hypointense T₁ lesion load over T₂ lesion load, this ratio being higher in patients with SPMS than those with RRMS. These results are in line with previous results from preliminary studies^5,6 and further support the hypothesis that hypointense lesions may be used as a putative marker to monitor fixed deficit in MS.

Since the ultimate goal of treatment in MS is to prevent accumulation of fixed deficit, it is important to evaluate which factors influence development of hypointense lesions. Our results show that degree of inflammatory activity, as depicted by Gd-enhancing lesions on monthly MR imaging, predicts only to some degree the increase in hypointense T₁ and hyperintense T₂ lesion load at long-term follow-up. Our long-term observations extend the correlation found between enhancing lesion rate and change in T₂ lesion load in recent studies^21,24; this correlation is slightly stronger than the correlation between new enhancing lesion rate and accumulation of hypointense lesions on T₁-weighted images. This suggests that not all enhancing lesions progress to hypointensity on T₁-weighted images, while most enhancing lesions will lead to lesions on T₂-weighted images. In a preliminary follow-up study investigating development and course of Gd-enhancing lesions,²⁵ 80% of enhancing lesions appeared hypointense on unenhanced T₁-weighted images. During follow-up (6 months), 44% of these "acute" hypointense lesions returned to isointensity, although they remained visible on T₂-weighted images, while 36% were persistent hypointense ("chronic"). Apparently, inflammatory activity leads to severe (irreversible) tissue destruction in only part of enhancing lesions, while other lesions are less severely damaged and capable of tissue repair. Since the pathogenesis—and subsequent clinical impact—might be different for acute and chronic hypointense lesions, they should be considered separately in follow-up studies. In our follow-up study, only postcontrast T₁-weighted images were available; therefore, a distinction between acute and chronic hypointense lesions was not possible. However, the contribution of acute hypointense lesions to our T₁ lesion load measurement has probably been small. First, most acute hypointense lesions return to isointensity within 3 months after enhancement²⁵; their impact in a long-term follow-up study will therefore be minimal. Second, our T₁-weighted images were performed after the administration of Gd. Since most acute hypointense lesions will show enhancement, they subsequently were not included in our T₁ lesion load measurements.

Our data suggest that inflammatory activity as expressed by Gd enhancement is only one of several factors related to development of destructive lesions, and that so far unknown factors may independently have a prominent role. This observation is in line with a number of clinical observations that relapse rate only shows moderate correlations with development of future disability. Further research should focus on the identification of these factors that, in addition to inflammatory activity, influence development of destructive and disabling lesions. Heterogeneity, with regard to pathological characteristics as well as to genetics of MS, may explain part of the observed variability. This hypothesis is in agreement with our observation that patients most likely to show an increase in hypointense T₁ lesion load are those who already have high hypointense T₁ lesion loads, thereby indicating that there is a subpopulation of patients with MS who are prone to develop destructive lesions. Hypointense "burden of disease" therefore could be used for stratification of patients at baseline in treatment trials where changes in T₁ lesion load will be analyzed as a secondary outcome measure. This shows strong similarity to the use of Gd-enhanced scanning to monitor new inflammatory activity, for which it was also shown that MR imaging activity at a single time point can predict future MR activity, suggesting that activity at baseline can be used for the stratification of patients for treatment trials.^21,24 Stratification of patients on the basis of hypointense lesion load implies that selection criteria for T₁ lesion load at study entry need to be developed. At this moment we do not have data available regarding this issue. However, before selection criteria can be determined, measurement of T₁ lesion load—to conform to measurement of T₂ lesion load—needs to be standardized between MR imaging centers. Further research is needed to develop guidelines for a clear definition of acute and chronic hypointense lesions, to optimize and standardize the pulse sequences used to identify hypointense lesions, and to improve intrarater and interrater variability in hypointense lesion load measurements.

In agreement with previous publications,^15,26,27 the influence of inflammatory activity on changes in clinical disability is difficult to establish, since only a positive correlation could be observed for patients with SPMS. The absence of a stronger correlation could be related to our relatively short follow-up period (approximately 40 months) and to the small number of patients with SPMS included (12 patients). Of these 12 patients with SPMS, only 6 showed an increase in EDSS score greater than 0.5 during follow-up. Most importantly, interobserver variation for EDSS assessment, which in our study was scored by 2 neurologists, may have resulted in loss of sensitivity.

Of technical concern in our follow-up study is the difference in field strength at study entry and study exit. Apart from increase in signal to noise, the relative sensitivity for detection of lesions probably varies between 0.6 T (first series of scans) and 1.5 T (last scan), since the amount of T₁ and T₂ weighting differs. The MR images at study entry and study exit therefore were analyzed comparatively for each patient. Further, because this difference in field strength may have influenced calculation of lesion volumes and with that our descriptive data, rank-order and cross-sectional correlations have been used to evaluate our data.

Despite limitations in clinical evaluations, (changes in) MR imaging variables, and duration of follow-up, it is clear that progressive disability only correlates weakly with presently analyzed MR variables. Therefore, a search for other laboratory markers is still needed. In terms of MR imaging variables, for example, cerebral atrophy was shown to be related to clinical worsening, even in the absence of new (enhancing) lesions.²⁸ Most likely a more general disease progress occurs "in between" focal lesions. Further research is warranted to characterize this diffuse process in brain^29-31 and spinal cord.³²

In conclusion, we have demonstrated that the degree of inflammatory activity as depicted by Gd-enhancing lesions is correlated to some degree with increase in hypointense lesion load at long-term follow-up, but other, so far unidentified, factors may have a prominent role. Initial hypointense T₁ lesion load strongly contributes to subsequent increase in T₁ lesion load, which indicates that there is a subpopulation of patients with MS who are prone to develop more destructive lesions.

Accepted for publication July 13, 1998.

Dr van Walderveen is funded by the Dutch Multiple Sclerosis Society, the Hague, the Netherlands (grant 95-245).

We thank Herman Ader, PhD, from the Department of Epidemiology and Biostatistics, Free University Amsterdam, Amsterdam, the Netherlands, for his assistance in analyzing the data, and Marlieke de Vos, MD, for helping with quantitation of brain abnormalities.

Reprints: Marianne A. A. van Walderveen, MD, Magnetic Resonance Center for Multiple Sclerosis Research, Academic Hospital Vrije Universiteit, PO Box 7057, 1007 MB Amsterdam, the Netherlands (e-mail: m.vanwalderveen@azvu.nl).