Early-stage volume losses in the corpus callosum and thalamus predict the progression of brain atrophy in patients with multiple sclerosis

doi:10.21203/rs.3.rs-2626591/v1

Download PDF

Article

Early-stage volume losses in the corpus callosum and thalamus predict the progression of brain atrophy in patients with multiple sclerosis

https://doi.org/10.21203/rs.3.rs-2626591/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

A method that can be used in the early stage of multiple sclerosis (MS) to predict the progression of brain volume loss (BVL) has not been fully established. Thus, to develop a method of predicting progressive BVL in patients with MS (pwMS), eighty-two consecutive Japanese pwMS—with either relapsing-remitting MS (86%) or secondary progressive MS (14%)—and 41 healthy controls were included in this longitudinal retrospective analysisover anobservational periodof approximately 3.5 years. Using a hierarchical cluster analysis with multivariate imaging data obtained by FreeSurfer analysis, we classified the pwMS into clusters. As a result, at baseline and follow-up, pwMS were cross-sectionally classified into three major clusters (Clusters 1, 2, and 3) in ascending order by disability and BVL. Among the patients included in Cluster 1 at baseline, approximately one-third of patients (12/52) transitioned into Cluster 2 at follow-up. The volumes of thecorpus callosum, the thalamus, and the whole brain excluding the ventricles were significantly decreased in the transition group compared with the nontransition group and were found to be the most important predictors of transition. Decreased volumes of thecorpus callosum and thalamus in the relatively early stage of MS may predict the development of BVL.

Health sciences/Neurology/Neurological disorders/Multiple sclerosis

Health sciences/Biomarkers/Prognostic markers

Multiple sclerosis (MS) is a chronic inflammatory demyelinating and neurodegenerative disease of the central nervous system ¹. The inflammatory component of MS pathology can be focal or diffuse and is associated with neurodegenerative processes that ultimately lead to irreversible tissue damage and neuronal loss. Neurodegeneration is recognized as being associated with acute inflammation in the early stages of MS and as the main driver of irreversible disability in this disease ². Since several recent studies have suggested that brain volume loss (BVL) can be ameliorated by escalation of disease-modifying therapies (DMTs) in patients with relapsing-remitting MS (RRMS) ³, there is currently a pressing need for clinicians to predict cases of future brain atrophy at an early disease stage.

The assessment of brain atrophy, particularly of the grey matter (GM), is of high clinical relevance because it is recognized as an independent predictor of current and future physical disability and cognitive impairment at the group level ^4,5. Additionally, several studies have suggested that atrophy of the corpus callosum can also be a strategic biomarker to monitor MS-related neurodegeneration ⁶. Recent studies have suggested that the disruption of white matter (WM) tracts impacts volume loss in connected deep GM over 5 years in patients with MS (pwMS) ⁷. However, it has not been elucidated whether atrophy of both the GM and WM tracts at an early disease stage can predict future brain atrophy.

In our previous study ⁸, based on a cluster analysis of global and regional brain volume data, we showed that the pattern of BVL in pwMS could be categorized into three main clusters (Clusters 1, 2, and 3), in ascending order by disability. The results showed that the volumes of the whole brain and the supratentorial regions, including the lateral ventricles, cerebral white matter, corpus callosum, cerebral cortex, subcortical grey matter, and thalamus, decreased sharply in Cluster 3 patients and moderately in Cluster 2 patients, whereas they were preserved in Cluster 1 patients and healthy controls (HCs) equally. Our study also indicated that different brain regions exhibit different degrees of atrophy against a background of global brain atrophy in MS.

Since we could cross-sectionally categorize pwMS into clusters with significantly different degrees of BVL that also reflected the heterogeneous trajectories of volume loss among different brain regions as described above ⁸, we used the same technique in this longitudinal study and attempted to find methods for predict cases of progressive BVL. We first investigated whether some patients who were classified into one cluster might transition to another cluster with more severe BVL during the clinical course. Furthermore, we tried to comprehensively clarify the characteristics of MS patients who would transition to a cluster with more severe BVL and find predictive factors regarding the transition among various clinical and radiological findings.

Patient clinical profiles

Eighty-two pwMS were included in this study. Demographic, clinical and radiological information on our study cohort at baseline and follow-up are presented in Tables 1 and 2 and Table 3, respectively. Compared with HCs, MS patients had significantly decreased brain volumes (Table 1). The study cohort consisted of patients presenting with CIS (1%), RRMS (85%), and secondary progressive MS (14%) at baseline (Table 2). The median observational period was 3.4 (IQR, 2.3–3.9) years. During follow-up, one CIS patient converted to RRMS, and 5 patients converted from RRMS to SPMS, based on the definition by Lublin et al. ⁹ (Table 3).

Table 1

Comparison between patients with multiple sclerosis (MS) and healthy controls (HCs).
	Total MS patients (n = 82)	HCs (n = 41)	p value	FDR-corrected p values
Sex (F, M)	n = 60 (73%), n = 22 (27%)	n = 27 (66%), n = 14 (34%)	0.4005	0.4806
Age, median (IQR)	39 (33–45)	41 (27–52)	0.6539	0.6539
BrainSegVol/eTICV, median (IQR)	0.720 (0.692–0.737)	0.742 (0.728–0.755)	< 0.0001	< 0.0003
TotalGrayVol/eTICV, median (IQR)	0.395 (0.380–0.412)	0.407 (0.393–0.423)	0.0032	0.0064
SubCortGrayVol/eTICV, median (IQR)	0.035 (0.032–0.036)	0.037 (0.036–0.039)	< 0.0001	< 0.0003
CortexVol/eTICV, median (IQR)	0.293 (0.280–0.304)	0.299 (0.289–0.317)	0.0153	0.02295
Data from MS patients are those at baseline. p values were evaluated by the Mann–Whitney U test. Bold font indicates FDR-corrected p values less than 0.05.

Table 2

Clinical profiles of MS patients at baseline.
	Total MS patients	Cluster 1	Cluster 2	Cluster 3	p values			FDR-corrected p values
	(n = 82)	(n = 52)	(n = 14)	(n = 16)	C3 vs. 1	C3 vs. 2	C2 vs. 1	C3 vs. 1	C3 vs. 2	C2 vs. 1
Sex (F, M)	n = 60, n = 22	n = 40, n = 12	n = 10, n = 4	n = 10, n = 6		0.5162			0.6309
Age*	39 (33–45)	39.5 (33-43.8)	34.5 (28.8–47.3)	42 (38-47.5)	0.2247	0.2269	0.7456	0.3317	0.3317	0.8484
Disease duration*	7.5 (4.2–13.3)	6.4 (2.5–10.7)	6.9 (4.5–10.6)	18 (10.1–24.0)	0.0002	0.0052	0.9218	0.0005	0.0101	0.9506
EDSS*	2 (1-3.5)	2 (1–3)	2 (1-3.1)	3.8 (2–6)	0.0027	0.0705	0.8474	0.0056	0.1193	0.9321
MSSS*	2.7 (0.9-5.0)	2.3 (0.9–4.6)	2.8 (1.1–4.7)	4.9 (0.91-8.0)	0.2312	0.4873	0.9546	0.3317	0.6309	0.9546
ARR, median (min-max)	0.26 (0-4.8)	0.23 (0-4.8)	0.45 (0–4)	0.4 (0.04–1.46)	0.378	0.9003	0.559	0.5198	0.9506	0.6588
BrainSegVol/eTICV*	0.720 (0.692–0.737)	0.732 (0.720–0.743)	0.704 (0.691–0.714)	0.647 (0.618–0.663)	< 0.0001	< 0.0001	< 0.0001	< 0.0003	< 0.0003	< 0.0003
TotalGrayVol/eTICV*	0.395 (0.380–0.412)	0.403 (0.391–0.417)	0.389 (0.382–0.403)	0.356 (0.334–0.370)	< 0.0001	0.0001	0.0583	< 0.0003	0.0003	0.1069
SubCortGrayVol/eTICV*	0.035 (0.032–0.036)	0.036 (0.035–0.037)	0.032 (0.032–0.033)	0.028 (0.026–0.029)	< 0.0001	< 0.0001	< 0.0001	< 0.0003	< 0.0003	< 0.0003
CortexVol/eTICV*	0.293 (0.280–0.304)	0.297 (0.288–0.309)	0.295 (0.286–0.299)	0.263 (0.252–0.277)	< 0.0001	0.0002	0.5008	< 0.0003	0.0005	0.6309
T2 lesion load (ml)*	4.49 (2.05–10.8)	2.93 (1.30–5.32)	9.37 (6.18–11.3)	21.6 (15.2–25.7)	< 0.0001	0.0004	0.0002	< 0.0003	0.0009	0.0005
Disease type
CIS	n = 1 (1%)	n = 1 (2%)	n = 0 (0%)	n = 0 (0%)		< 0.0001			< 0.0003
RRMS	n = 70 (85%)	n = 50 (96%)	n = 13 (93%)	n = 7 (44%)
SPMS	n = 11 (14%)	n = 1 (2%)	n = 1 (7%)	n = 9 (56%)
DMT						0.0723			0.1193
High-efficacy	n = 41 (50%)	n = 23 (44%)	n = 5 (36%)	n = 13 (81%)
Baseline	n = 34 (41%)	n=༒༔ (46%)	n = 8 (57%)	n = 2 (13%)
None	n = 7 (9%)	n = 5 (10%)	n = 1 (7%)	n = 1 (6%)
p values were evaluated by the Steel-Dwass test. Bold font indicates FDR-corrected p values less than 0.05. * Variables are presented as the median (interquartile range).
Abbreviations: C1, Cluster 1; C2, Cluster 2; C3, Cluster 3; high-efficacy, high-efficacy DMTs (fingolimod and natalizumab); baseline, baseline DMTs (interferon beta and dimethyl fumarate).

Table 3

Clinical profiles of MS patients at follow-up.
	Total MS patients	Cluster 1	Cluster 2	Cluster 3	p values			FDR-corrected p values
	(n = 82)	(n = 40)	(n = 26)	(n = 16)	C3 vs. 1	C3 vs. 2	C2 vs. 1	C3 vs. 1	C3 vs. 2	C2 vs. 1
Sex (F, M)	n = 60, n = 22	n = 33, n = 7	n = 17, n = 9	n = 10, n = 6	0.1734			0.2772
Age*	42 (36–48)	42.5 (37.3–48.8)	38.5 (31–46)	45.5 (40.5–49.8)	0.3809	0.0565	0.1611	0.5161	0.1032	0.2706
Disease duration (y)*	11 (7.4–16.5)	9.5 (6.4–15.0)	8.4 (5.5–12.2)	21.1 (12.7–27.1)	0.0018	0.0003	0.6782	0.0044	0.0008	0.7698
EDSS*	2 (0.75–3.5)	1.5 (0–2)	2 (0.75–3.5)	5 (2-6.75)	0.0011	0.0256	0.4612	0.0029	0.0538	0.5534
MSSS*	1.77 (0.43–4.55)	1.28 (0.28–3.11)	1.54 (0.82–4.57)	4.62 (0.75–7.40)	0.0419	0.4424	0.3734	0.0838	0.5534	0.5161
BrainSegVol/eTICV*	0.713 (0.689–0.733)	0.733 (0.721–0.746)	0.701 (0.691–0.712)	0.644 (0.618–0.656)	< 0.0001	< 0.0001	< 0.0001	< 0.0003	< 0.0003	< 0.0003
TotalGrayVol/eTICV*	0.392 (0.376–0.406)	0.400 (0.388–0.414)	0.392 (0.378–0.397)	0.349 (0.335–0.368)	< 0.0001	< 0.0001	0.063	< 0.0003	< 0.0003	0.1103
SubCortGrayVol/eTICV*	0.034 (0.031–0.036)	0.036 (0.035–0.038)	0.033 (0.031–0.034)	0.028 (0.027–0.029)	< 0.0001	< 0.0001	< 0.0001	< 0.0003	< 0.0003	< 0.0003
CortexVol/eTICV*	0.289 (0.280–0.301)	0.295 (0.285–0.309)	0.290 (0.285–0.297)	0.259 (0.248–0.275)	< 0.0001	< 0.0001	0.4068	< 0.0003	< 0.0003	0.5339
T2 lesion load (ml)*	5.45 (2.17–12.2)	2.23 (1.06–4.09)	7.21 (3.91–11.2)	21.0 (14.2–24.5)	< 0.0001	< 0.0001	< 0.0001	< 0.0003	< 0.0003	< 0.0003
Disease type
CIS > RRMS	n = 1 (1%)	n = 1 (2.5%)	n = 0 (0%)	n = 0 (0%)	< 0.0001			< 0.0003
RRMS > RRMS	n = 65 (79%)	n = 37 (92.5%)	n = 23 (88%)	n = 5 (31%)
RRMS > SPMS	n = 5 (6%)	n = 1 (2.5%)	n = 2 (8%)	n = 2 (13%)
SPMS > SPMS	n = 11 (13%)	n = 1 (2.5%)	n = 1 (4%)	n = 9 (56%)
DMT
High-efficacy	n = 36 (44%)	n = 16 (40%)	n = 9 (35%)	n = 11 (69%)	0.1782			0.2772
Baseline	n = 23 (28%)	n = 14 (35%)	n = 9 (35%)	n = 0 (0%)
None	n = 6 (7%)	n = 3 (8%)	n = 2 (8%)	n = 1 (6%)
Switch	n = 17 (21%)	n = 7 (18%)	n = 6 (23%)	n = 4 (25%)
Observational period (y)*	3.38 (2.33–3.92)	3.38 (1.94–3.83)	3.29 (2.73-4)	3.54 (2.25–3.92)	0.9005	0.967	0.7302	0.9608	0.9906	0.8071
ARR (observational period) **	0 (0-0.97)	0 (0-0.75)	0 (0-0.97)	0 (0-0.63)	0.9922	0.476	0.2754	0.9922	0.5553	0.3989
ARR (disease course) **	0.23 (0-1.49)	0.20 (0-0.63)	0.24 (0-1.49)	0.35 (0.04–1.22)	0.0471	0.915	0.2026	0.0899	0.9608	0.3039
	Total MS (n = 72)	Cluster 1 (n = 36)	Cluster 2 (n = 22)	Cluster 3 (n = 14)	C3 vs. 1	C3 vs. 2	C2 vs. 1	C3 vs. 1	C3 vs. 2	C2 vs. 1
Processing speed (Z score)	-1.04 (-1.89–0.21)	-0.68 (-1.54-0.12)	-0.94 (-2.13–0.41)	-2.48 (-4.28–1.11)	0.0028	0.0203	0.4568	0.0065	0.0449	0.5534
p values were evaluated by the Steel-Dwass test. Bold font indicates FDR-corrected p values less than 0.05. Median (interquartile range), *Median (min-max).
Abbreviations: C1, Cluster 1; C2, Cluster 2; C3, Cluster 3; CIS > RRMS, CIS patients converted to RRMS patients; RRMS > RRMS, RRMS patients who remained as RRMS patients at follow-up; RRMS > SPMS, RRMS patients converted to SPMS patients; SPMS > SPMS, SPMS patients who remained as SPMS patients at follow-up; high-efficacy, high-efficacy DMTs (fingolimod and natalizumab); baseline, baseline DMTs (interferon beta and dimethyl fumarate); switch, DMT was switched from one to another.

Cluster analysis identified three clusters based on global and regional brain volume data

The cluster analysis results at baseline and follow-up are shown as dendrograms (Fig. 1). Each cluster analysis identified three clusters (Clusters 1, 2, and 3). Patients who were classified into Clusters 1, 2, and 3 at baseline are represented by red, green, and blue colors, respectively, in the two dendrograms. When we compared the two dendrograms, 12 cases initially included in Cluster 1 (cases in red boxes in Fig. 1) transitioned to Cluster 2 at follow-up. Interestingly, among the 12 transitioned cases, 8 and 3 cases formed their own small clusters at baseline.

Comparisons of clinical and radiological findings among the three clusters at baseline

At baseline, the 82 MS patients were classified into 3 groups based on MRI volumetric data in the cluster analysis, with 52, 14, and 16 patients grouped into Clusters 1, 2, and 3, respectively (Table 2). Patient age and sex did not significantly differ among the 3 clusters. Disease duration was significantly longer in Cluster 3 than in Cluster 1 and Cluster 2. EDSS scores were significantly higher in Cluster 3 than in Cluster 1. Cluster 1 consisted of patients with CIS (2%), RRMS (96%), and SPMS (2%). Cluster 2 consisted of patients with RRMS (93%) and SPMS (7%). Cluster 3 consisted of patients with RRMS (44%) and SPMS (56%). The chi-squared test revealed that there was a significant difference in the proportions of these subtypes among the three clusters. Whole-brain volume and subcortical GM volume were significantly different among the three clusters and decreased in the order of Cluster 3, 2, and 1. Lesion load was significantly different among the three clusters and increased in the order of Clusters 3, 2, and 1. Total GM volume and cortical GM volume were significantly decreased in Cluster 3 compared with the volumes in Clusters 1 and 2.

Comparisons of clinical and radiological findings among the three clusters at follow-up

At follow-up, the 82 MS patients were classified into 3 groups in the cluster analysis, with 40, 26, and 16 patients grouped into Clusters 1, 2, and 3, respectively (Table 3). The patterns of significant differences among the three clusters at follow-up were similar to those at baseline (Tables 2 and 3). Disease duration was significantly longer in Cluster 3 than in Clusters 1 and 2. EDSS scores were significantly higher in Cluster 3 than in Cluster 1. The proportions of disease subtype were significantly different among the three clusters. Whole-brain volume and subcortical GM volume were significantly different among the three clusters and decreased in the order of Cluster 3, 2, and 1. Lesion load was significantly different among the three clusters and increased in the order of Clusters 3, 2, and 1. Total GM volume and cortical GM volume were significantly decreased in Cluster 3 compared with the volumes in Clusters 1 and 2. IPS was evaluated in 72 MS patients, while Clusters 1, 2, and 3 included 36, 22, and 14 patients, respectively. Z scores of IPS were significantly decreased in Cluster 3 compared with those in Clusters 1 and 2.

Comparison of baseline data between transition and nontransition groups

Among the 52 patients who were initially classified into Cluster 1 at baseline, 12 patients transitioned to Cluster 2, and the other 40 patients remained in Cluster 1 at follow-up (Fig. 1). Therefore, we defined the former and latter groups as transition (CL1 > CL2) and nontransition groups (CL1 > CL1), respectively. Clinical data, such as age, sex, disease duration, EDSS, MSSS, ARR, DMT usage, and disease subtypes, were not significantly different between the transition and nontransition groups (Table 4 − 1). However, the regional volumes of the corpus callosum, thalamus, and pallidum were significantly decreased in the transition group compared with those in the nontransition group (Fig. 2A-C). Whole-brain (Fig. 2D), supratentorial brain, cerebral WM, and subcortical GM volumes were also significantly decreased in the transition group compared with the nontransition group. The volume of WM hypointensity and lesion loads (Fig. 2E) were significantly increased in the transition group compared with those in the nontransition group (Table 4 − 2).

Table 4-1

Comparison of baseline data between transition and nontransition groups.
	CL1 > CL1 (n = 40)	CL1 > CL2 (n = 12)	p value	FDR-p
Age (y)	39.3 ± 7.38	34.1 ± 7.25	0.0542	0.1275
Sex (F, M)	33 (83%), 7 (17%)	7 (58%), 5 (42%)	0.0814	0.1642
Duration (y)	6.88 (3.73–11.4)	3.67 (0.83–8.42)	0.0842	0.1642
EDSS	2 (1–3)	1 (0-2.75)	0.1344	0.2444
MSSS	2.52 (0.96–4.48)	1.11 (0.43–5.61)	0.46	0.5576
ARR	0.22 (0.13–0.42)	0.28 (0.19–1.19)	0.2628	0.4205
DMT			0.9039	0.96540
High-efficacy	16 (40%)	5 (42%)
Baseline	14 (35%)	3 (25%)
None	3 (7.5%)	1 (8%)
Switch	7 (17.5%)	3 (25%)
Disease course			0.8121	0.9023
CIS > RRMS	1 (2.5%)	0 (0%)
RRMS > RRMS	37 (92.5%)	12 (100%)
RRMS > SPMS	1 (2.5%)	0 (0%)
SPMS > SPMS	1 (2.5%)	0 (0%)

Table 4-2

Comparison of baseline data between transition and nontransition groups.
	CL1 > CL1 (n = 40)	CL1 > CL2 (n = 12)	p value	FDR-p
CC/eTICV	0.0021 (0.0019–0.0023)	0.0017 (0.0017–0.0018)	< 0.0001	< 0.002
Thalamus/eTICV	0.0093 (0.0088–0.0096)	0.0083 (0.0080–0.0088)	0.0001	0.0020
BrainSegVolNotVent/eTICV	0.7227 (0.7117–0.7386)	0.7030 (0.6904–0.7094)	0.0002	0.0020
SupraTentorialVolNotVent/eTICV	0.6361 (0.6239–0.6516)	0.6160 (0.6088–0.6237)	0.0002	0.0020
BrainSegVol/eTICV	0.7380 (0.7261–0.7525)	0.7180 (0.7087–0.7280)	0.0003	0.0024
SupraTentorialVol/eTICV	0.6499 (0.6399–0.6650)	0.6313 (0.6252–0.6418)	0.0006	0.0040
CerebralWhiteMatterVol/eTICV	0.3022 (0.2911–0.3102)	0.2844 (0.2774–0.2951)	0.001	0.0057
Pallidum/eTICV	0.0025 (0.0023–0.0026)	0.0023 (0.0020–0.0023)	0.002	0.0100
WM-hypointensities/eTICV	0.0011 (0.0008–0.0016)	0.0018 (0.0014–0.0025)	0.0041	0.0171
SubCortGrayVol/eTICV	0.0363 (0.0355–0.0382)	0.0352 (0.0338–0.0360)	0.0044	0.0171
Lesion load (ml)	2.26 (0.93–3.93)	5.15 (2.92–8.53)	0.0047	0.0171
Choroid-plexus/eTICV	0.0008 (0.0007–0.0009)	0.0010 (0.0008–0.0010)	0.0054	0.0180
3rd -Ventricle/eTICV	0.0007 (0.0006–0.0009)	0.0009 (0.0008–0.0012)	0.0118	0.0337
Optic-Chiasm/eTICV	0.0001 (0.0001–0.0001)	0.0001 (0.0001–0.0002)	0.0118	0.0337
Putamen/eTICV	0.0065 (0.0060–0.0068)	0.0061 (0.0058–0.0063)	0.0213	0.0568
Accumbens-area/eTICV	0.0007 (0.0006–0.0007)	0.0006 (0.0006–0.0007)	0.0371	0.0928
Lateral-Ventricle/eTICV	0.0106 (0.0082–0.0135)	0.0120 (0.0111–0.0155)	0.0714	0.1587
Inf-Lat-Vent/eTICV	0.0006 (0.0004–0.0008)	0.0008 (0.0006–0.0011)	0.0862	0.1642
Non-WM-hypointensities/eTICV	0.0000 (0.0000–0.0000)	0.0000 (0.0000–0.0000)	0.1585	0.2757
Cerebellum-White-Matter/eTICV	0.0184 (0.0172–0.0194)	0.0176 (0.0164–0.0188)	0.2587	0.4205
TotalGrayVol/eTICV	0.4042 (0.3917–0.4174)	0.3958 (0.3903–0.4079)	0.2872	0.4418
Hippocampus/eTICV	0.0053 (0.0050–0.0055)	0.0054 (0.0051–0.0058)	0.3178	0.4521
SurfaceHoles/eTICV	0.0000 (0.0000–0.0000)	0.0000 (0.0000–0.0000)	0.3178	0.4521
Brain-Stem/eTICV	0.0138 (0.0129–0.0145)	0.0137 (0.0131–0.0139)	0.3393	0.4521
Vessel/eTICV	0.0000 (0.0000-0.0001)	0.0001 (0.0000-0.0001)	0.3393	0.4521
Caudate/eTICV	0.0045 (0.0042–0.0047)	0.0044 (0.0041–0.0046)	0.3504	0.4521
CSF/eTICV	0.0007 (0.0006–0.0009)	0.0008 (0.0007–0.0008)	0.4092	0.5115
CortexVol/eTICV	0.2980 (0.2882–0.3106)	0.2943 (0.2820–0.3067)	0.4871	0.5731
Amygdala/eTICV	0.0021 (0.0020–0.0022)	0.0022 (0.0019–0.0023)	0.5431	0.6207
4th -Ventricle/eTICV	0.0011 (0.0010–0.0013)	0.0011 (0.0010–0.0013)	0.9308	0.9654
VentralDC/eTICV	0.0050 (0.0049–0.0051)	0.0050 (0.0048–0.0053)	0.948	0.9654
Cerebellum-Cortex/eTICV	0.0683 (0.0638–0.0729)	0.0671 (0.0648–0.0722)	0.9654	0.9654
P values were evaluated by the Mann–Whitney U test. Bold font indicates FDR-corrected p values less than 0.05.
Abbreviations: CL1 > CL1, patients who were classified into cluster 1 at baseline and remained in cluster 1 at follow-up; CL1 > CL2, patients who were classified into cluster 1 at baseline and transitioned to cluster 2 at follow-up.

Predictor Screening

To determine which of the clinical and radiological findings at baseline are important to determine whether MS patients in Cluster 1 would transition into Cluster 2, we performed predictor screening. Table 5 shows the list of the top 20 predictors with their respective contributions. Predictors with the highest contributions are likely to be important in predicting outcome. The volumes of the corpus callosum, pallidum, thalamus, and whole brain excluding ventricles were found to be the most important predictors.

Table 5

Results of predictor screening
Predictor	Contribution	Portion
CC/eTICV	10.4364	0.3628
BrainSegVolNotVent/eTICV	2.8501	0.0991
Pallidum/eTICV	2.2783	0.0792
Thalamus/eTICV	1.9785	0.0688
Optic-Chiasm/eTICV	1.1414	0.0397
Age	1.1403	0.0396
BrainSegVol/eTICV	1.0487	0.0365
Putamen/eTICV	0.9128	0.0317
SupraTentorialVol/eTICV	0.8013	0.0279
SupraTentorialVolNotVent/eTICV	0.586	0.0204
Choroid-plexus/eTICV	0.4855	0.0169
Lesion load	0.4215	0.0147
WM-hypointensities/eTICV	0.3898	0.0136
ARR	0.3775	0.0131
3rd-Ventricle/eTICV	0.3478	0.0121
Cerebellum-Cortex/eTICV	0.3146	0.0109
CortexVol/eTICV	0.2856	0.0099
EDSS	0.2843	0.0099
MSSS	0.2781	0.0097
Vessel/eTICV	0.2769	0.0096
The top 20 predictors with the highest contribution to determining whether MS patients in Cluster 1 would transition to Cluster 2 are listed. Predictors with the highest contributions are likely to be important in predicting outcomes. The Portion column in the table shows the percent contribution of each variable.

In this study, we cross-sectionally classified MS patients into three major clusters (Clusters 1, 2, and 3) that showed increasing disability and BVL in that order at baseline and follow-up evaluated by the same method. Furthermore, in this longitudinal retrospective analysis over the approximate observational period of 3.5 years, among the patients included in Cluster 1 at baseline, approximately one-third of patients transitioned into Cluster 2 at follow-up. The volumes of the corpus callosum, thalamus, and whole brain excluding ventricles were significantly decreased in the transition group compared with those in the nontransition group and were found to be the most important predictors of transition among various clinical and radiological findings.

The speed of BVL in MS patients differs substantially among brain regions and among different phases of the disease ^10,11. This indicates that a simple assessment of the volume of the whole brain is not sufficient to precisely capture the progression of the brain pathology in MS patients. Thus, MS brain pathology staging should consider the heterogeneous trajectories of volume loss among different brain regions. To date, several studies have suggested that data-driven approaches suitable for processing multidimensional data may be a powerful tool for staging complex MS brain pathology ^8,10,11. A recent study by Miyazaki et al. also performed cluster analysis using global and regional brain volume data that were obtained by FreeSurfer analysis. Cluster analysis of the global and regional brain volume data resulted in patients being classified into three clusters (Clusters 1, 2, and 3) in order of decreasing global brain volume. Furthermore, SDMT scores differed significantly among the three clusters, with a decrease from Clusters 1 to 3.

In this study, the decreased volume of the corpus callosum was found to be one of the most important predictors of the transition between clusters. The corpus callosum is a dense WM structure connecting the cerebral hemispheres ⁶. While the corpus callosum is relatively stable in the normal ageing population, it exhibits significant atrophy in MS patients ¹². Several cross-sectional and longitudinal studies, including ours, have shown that callosal volume loss/atrophy is significantly associated with cognitive and physical impairment in MS patients ^13–16. Moreover, the corpus callosum is a region that can be easily evaluated for its volume by both 2D and 3D MRI measurements^14–16. This study indicates that corpus callosum atrophy can be a strategic biomarker to monitor MS-related neurodegeneration in the early stage, as suggested in previous studies ⁶.

The decreased volume of the thalamus was also found to be one of the most important predictors for the transition between clusters. The thalamus represents a relay station of cortico-subcortical and cortico-cortical networks ¹⁷ and is involved in a wide range of neurologic functions, including motor, sensory, integrative, and higher cortical functions ¹⁸. Volume loss/atrophy of the thalamus has been well documented in MS patients using MRI and is clinically relevant ^19–21. Pathologically, focal thalamic lesions in MS patients exhibit neuronal loss, demyelination, and axonal transection ²¹. Moreover, thalamic volume loss is observed on MRI in the earliest identifiable phases of MS, including pediatric MS, CIS, and radiologically isolated syndrome, whereas whole brain volume and total GM volume may still be preserved ^21,22. Location, unique functions, and vulnerability to MS neuropathology from the earliest disease stages render the thalamus a critical barometer of diffuse neuropathologic damage in MS patients ¹⁸. To date, thalamic atrophy has been associated with the evolution of definite MS and disability progression in early RRMS patients and with the evolution of PPMS ^23–25. Therefore, our results were basically in accordance with these previous studies.

Our study has several limitations. First, this investigation was a single-centre study performed with a limited sample of Japanese MS patients; therefore, the results might have been influenced by selection bias. Furthermore, this categorization scheme may not be appropriate for other cohorts, since Japanese MS patients may have a slightly milder clinical course than Caucasian patients ²⁶. However, there is general agreement that the Western-type MS observed in patients in Asia is not fundamentally different from that observed in typical MS patients of the Caucasian population once NMO and NMOSDs have been excluded ²⁷. Second, we could not observe whether some patients who were classified into Cluster 2 might transition into Cluster 3 in this study. In the future, a multicentered more longitudinal dataset is needed to perform a complete assessment of these measures and to obtain generalizability of the results.

In conclusion, we found that approximately one-third of patients in the mild BVL group transitioned to the moderate BVL group over the observational period of approximately 3.5 years. Furthermore, we showed that decreased volumes of the corpus callosum, the thalamus, and the whole brain excluding the ventricles may predict the future development of brain atrophy in the relatively early disease stage. Atrophy of the corpus callosum, a major WM tract, and of the thalamus, a GM relay station, might constitute neurodegenerative findings that start in the early stage of MS and lead to further BVL in the future.

Patients

Eighty-two consecutive Japanese MS patients and 41 HCs who visited the Department of Neurology at Tohoku Medical and Pharmaceutical University Hospital, Sendai, Japan, between 2017 and 2021 were recruited retrospectively. The inclusion criteria were as follows: 1) MS patients and/or clinically isolated syndrome (CIS) patients as defined by the 2017 revision of the McDonald criteria ²⁸ and 2) age between 20 and 70 years. The exclusion criteria were as follows: 1) neuromyelitis optica spectrum disorders (NMOSDs) or myelin oligodendrocyte glycoprotein antibody-associated disorders (MOGADs), 2) a history of psychiatric illness other than stable depressive symptoms, and 3) a follow-up period of less than half a year. We used the Expanded Disability Status Scale (EDSS) ²⁹ and Multiple Sclerosis Severity Score (MSSS) ³⁰ to measure the patients’ disability and severity of illness, respectively. Among the 82 MS patients, 72 who agreed to be evaluated for their information processing speed (IPS) had undergone cognitive assessments with CogEval (Biogen Inc.) (https://apps.apple.com/us/app/cogeval/id1366437045). CogEval is an iPad-based screening assessment tool designed to evaluate cognitive function in pwMS and is based on and validated against the Symbol Digit Modalities Test (SDMT) ^31,32. The local institutional Ethics Committee at Tohoku Medical and Pharmaceutical University approved the study protocol (2017-2-011). The study was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Written informed consent was obtained from all participants.

MRI acquisition

All study subjects were scanned on the same whole-body 1.5 Tesla MRI system (MAGNETOM Aera, Siemens, Germany). The MR acquisition protocol included a high-resolution sagittal 3-dimensional (3D) T1-weighted magnetization-prepared rapid gradient-echo (MPRAGE) sequence (repetition time (TR): 2730 ms; echo time (TE): 3.3 ms; inversion time (TI): 1000 ms; 176 slices; field of view (FoV): 256 mm; and measured isotropic voxel size: 1×1×1 mm) and a sagittal 3D fluid-attenuated inversion recovery (FLAIR) sequence (TR: 5000 ms; TE: 335 ms; TI: 1800 ms; 176 slices; FoV: 256 mm; and measured isotropic voxel size: 1×1×1 mm).

MRI postprocessing to measure global and regional brain volumes

In brief, the 3D-MPRAGE dataset served as the input data for the postprocessing pipeline, and the regional and whole-brain volumes were estimated using the automated FreeSurfer stream (version 7.1.0; https://surfer.nmr.mgh.harvard.edu/), as previously described ^8,14,15,33. The brain volume data for a total of 63 segments were directly extracted from FreeSurfer’s automated segmentation results for each patient and normalized to their head size using total intracranial volume. These unitless values were used for further analysis.

In cluster analysis, we used the unitless values of the 63 regions of interest, as previously described ¹⁵. However, in the comparison of brain volumes among clusters or predictor screening, we used summed unitless values, i.e., summed values of the left and right thalamus or those of the anterior, middle anterior, central, middle posterior, and posterior portions of the corpus callosum.

Lesion volumetry by icobrain ms

The 3D T1 MPRAGE and 3D FLAIR datasets obtained from each MS patient were analyzed using the program “icobrain ms” by uploading the DICOM data to the Icometrix website (http://icometrix.com) as previously described ^8,14,33.

Cluster analysis of regional and global brain volumes

We performed a hierarchical cluster analysis using brain MRI volume data to categorize the MS patients into several subtypes, each consisting of patients who shared a similar pattern of brain volume, as previously described ^8,14,15,33. The cluster analysis was performed using the 63 regions of interest in the brain volumes obtained in each of the 82 subjects with MS.

Statistical analysis

Statistical analyses were performed using JMP Pro version 16.2.0 software. Quantitative variables are presented as the median (interquartile range (IQR)/minimum and maximum). Comparisons of numerical variables between disease groups were performed by the Mann–Whitney U test or the Steel-Dwass test, which controls for the overall experimentwise error rate. Comparisons of categorical variables were performed by chi-square tests. To screen many predictors for their ability to predict an outcome, we performed predictor screening using bootstrap forest partitioning to evaluate the contribution of predictors to the response. Predictor screening can identify predictors that might be weak when used alone but strong when used in combination with other predictors. P values were corrected for the false-discovery rate (FDR) using the Benjamini–Hochberg method ³⁴ (to account for type I error due to multiple comparisons), and results with FDR p values (P values adjusted for the FDR) less than 0.05 were considered statistically significant.

ARR, annualized relapse rate; BrainSegVol, brain segmentation volume; BrainSegVolNotVent, brain segmentation volume without ventricles; CC, corpus callosum; CIS, clinically isolated syndrome; CortexVol, cortical grey matter volume; CSF, cerebrospinal fluid; DMT, disease-modifying therapy; EDSS, Expanded Disability Status Scale; eTICV: estimated total intracranial volume; FDR, false-discovery rate; HCs, healthy controls; Inf-Lat-Vent, inferior lateral ventricle; IQR, interquartile range; MSSS, Multiple Sclerosis Severity Score; RRMS, relapsing-remitting MS; SPMS, secondary progressive MS; SubCortGrayVol, subcortical grey matter volume; SupraTentorialVol: supratentorial volume; SupraTentorialVolNotVent: supratentorial volume without ventricles; SurfaceHoles, total number of defect holes on the surfaces prior to fixing; TotalGrayVol, total grey matter volume; VentralDC, ventral diencephalon; WM, white matter.

Data availability

The datasets used and analyzed during the current study is available from the corresponding author on reasonable request.

Acknowledgements

IN is receiving research support from LSI Medience and is funded by MHLW Program Grant Number 20FC1030 and JSPS KAKENHI Grant Number 20K07892. JF have no disclosures to report. We thank Sorama Aoki for supporting statistical analysis.

Author contributions

JF contributed to the study conception, design, data collection, and analysis. All authors reviewed and approved the final manuscript.

Additional information

Conflicts of interest

IN is serving on scientific advisory boards for Biogen Japan and Novartis Pharma and is receiving honouraria for speaking engagements with Biogen Japan, Mitsubishi Tanabe Pharma, Novartis Pharma, Takeda Pharmaceutical, and Eisai. JF have no conflicts of interest to disclose.

Reich, D. S., Lucchinetti, C. F. & Calabresi, P. A. Multiple Sclerosis. The New England journal of medicine 378, 169-180, doi:10.1056/NEJMra1401483 (2018).
Sastre-Garriga, J. et al. MAGNIMS consensus recommendations on the use of brain and spinal cord atrophy measures in clinical practice. Nature reviews. Neurology 16, 171-182, doi:10.1038/s41582-020-0314-x (2020).
Uher, T. et al. Evolution of Brain Volume Loss Rates in Early Stages of Multiple Sclerosis. Neurology(R) neuroimmunology & neuroinflammation 8, doi:10.1212/nxi.0000000000000979 (2021).
Filippi, M. et al. Diagnosis of Progressive Multiple Sclerosis From the Imaging Perspective: A Review. JAMA neurology 78, 351-364, doi:10.1001/jamaneurol.2020.4689 (2021).
Van Schependom, J., Guldolf, K., D'Hooghe M, B., Nagels, G. & D'Haeseleer, M. Detecting neurodegenerative pathology in multiple sclerosis before irreversible brain tissue loss sets in. Transl Neurodegener 8, 37, doi:10.1186/s40035-019-0178-4 (2019).
Platten, M. et al. Deep Learning Corpus Callosum Segmentation as a Neurodegenerative Marker in Multiple Sclerosis. Journal of neuroimaging : official journal of the American Society of Neuroimaging 31, 493-500, doi:10.1111/jon.12838 (2021).
Carolus, K. et al. Time course of lesion-induced atrophy in multiple sclerosis. Journal of neurology 269, 4478-4487, doi:10.1007/s00415-022-11094-y (2022).
Fujimori, J. et al. Patterns of regional brain volume loss in multiple sclerosis: a cluster analysis. Journal of neurology 267, 395-405, doi:10.1007/s00415-019-09595-4 (2020).
Lublin, F. D. et al. Defining the clinical course of multiple sclerosis: the 2013 revisions. Neurology 83, 278-286, doi:10.1212/wnl.0000000000000560 (2014).
Eshaghi, A. et al. Progression of regional grey matter atrophy in multiple sclerosis. Brain : a journal of neurology 141, 1665-1677, doi:10.1093/brain/awy088 (2018).
Miyazaki, Y. et al. Stages of brain volume loss and performance in the Brief International Cognitive Assessment for Multiple Sclerosis. Multiple sclerosis and related disorders 67, 104183, doi:10.1016/j.msard.2022.104183 (2022).
Martola, J. et al. Progression of non-age-related callosal brain atrophy in multiple sclerosis: a 9-year longitudinal MRI study representing four decades of disease development. Journal of neurology, neurosurgery, and psychiatry 78, 375-380, doi:10.1136/jnnp.2006.106690 (2007).
Granberg, T. et al. Corpus callosum atrophy is strongly associated with cognitive impairment in multiple sclerosis: Results of a 17-year longitudinal study. Multiple sclerosis (Houndmills, Basingstoke, England) 21, 1151-1158, doi:10.1177/1352458514560928 (2015).
Fujimori, J. et al. Measurements of the corpus callosum index and fractional anisotropy of the corpus callosum and their cutoff values are useful to assess global brain volume loss in multiple sclerosis. Multiple sclerosis and related disorders 45, 102388, doi:10.1016/j.msard.2020.102388 (2020).
Nishizawa, K., Fujimori, J. & Nakashima, I. Two-dimensional measurements with cut-off values are useful for assessing brain volume, physical disability, and processing speed in multiple sclerosis. Multiple sclerosis and related disorders 59, 103543, doi:10.1016/j.msard.2022.103543 (2022).
Ajitomi, S., Fujimori, J. & Nakashima, I. Usefulness of two-dimensional measurements for the evaluation of brain volume and disability in multiple sclerosis. Multiple sclerosis journal - experimental, translational and clinical 8, 20552173211070749, doi:10.1177/20552173211070749 (2022).
Di Filippo, M., Portaccio, E., Mancini, A. & Calabresi, P. Multiple sclerosis and cognition: synaptic failure and network dysfunction. Nature reviews. Neuroscience 19, 599-609, doi:10.1038/s41583-018-0053-9 (2018).
Minagar, A. et al. The thalamus and multiple sclerosis: modern views on pathologic, imaging, and clinical aspects. Neurology 80, 210-219, doi:10.1212/WNL.0b013e31827b910b (2013).
Houtchens, M. K. et al. Thalamic atrophy and cognition in multiple sclerosis. Neurology 69, 1213-1223, doi:10.1212/01.wnl.0000276992.17011.b5 (2007).
Azevedo, C. J. et al. Thalamic atrophy in multiple sclerosis: A magnetic resonance imaging marker of neurodegeneration throughout disease. Annals of neurology 83, 223-234, doi:10.1002/ana.25150 (2018).
Ontaneda, D. et al. Deep grey matter injury in multiple sclerosis: a NAIMS consensus statement. Brain : a journal of neurology 144, 1974-1984, doi:10.1093/brain/awab132 (2021).
Henry, R. G. et al. Regional grey matter atrophy in clinically isolated syndromes at presentation. Journal of neurology, neurosurgery, and psychiatry 79, 1236-1244, doi:10.1136/jnnp.2007.134825 (2008).
Mesaros, S. et al. Thalamic damage predicts the evolution of primary-progressive multiple sclerosis at 5 years. AJNR. American journal of neuroradiology 32, 1016-1020, doi:10.3174/ajnr.A2430 (2011).
Zivadinov, R. et al. Thalamic atrophy is associated with development of clinically definite multiple sclerosis. Radiology 268, 831-841, doi:10.1148/radiol.13122424 (2013).
Niiranen, M. et al. Grey matter atrophy in patients with benign multiple sclerosis. Brain and behavior 12, e2679, doi:10.1002/brb3.2679 (2022).
Piccolo, L. et al. Multiple sclerosis in Japan appears to be a milder disease compared to the UK. Journal of neurology 262, 831-836, doi:10.1007/s00415-015-7637-3 (2015).
Polman, C. H. et al. Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria. Annals of neurology 69, 292-302, doi:10.1002/ana.22366 (2011).
Thompson, A. J. et al. Diagnosis of multiple sclerosis: 2017 revisions of the McDonald criteria. The Lancet. Neurology 17, 162-173, doi:10.1016/s1474-4422(17)30470-2 (2018).
Kurtzke, J. F. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology 33, 1444-1452, doi:10.1212/wnl.33.11.1444 (1983).
Roxburgh, R. H. et al. Multiple Sclerosis Severity Score: using disability and disease duration to rate disease severity. Neurology 64, 1144-1151, doi:10.1212/01.wnl.0000156155.19270.f8 (2005).
Rao, S. M. et al. Processing speed test: Validation of a self-administered, iPad(®)-based tool for screening cognitive dysfunction in a clinic setting. Multiple sclerosis (Houndmills, Basingstoke, England) 23, 1929-1937, doi:10.1177/1352458516688955 (2017).
Kalb, R. et al. Recommendations for cognitive screening and management in multiple sclerosis care. Multiple sclerosis (Houndmills, Basingstoke, England) 24, 1665-1680, doi:10.1177/1352458518803785 (2018).
Fujimori, J., Fujihara, K., Wattjes, M. & Nakashima, I. Patterns of cortical grey matter thickness reduction in multiple sclerosis. Brain and behavior 11, e02050, doi:10.1002/brb3.2050 (2021).
Benjamini, Y. & Hochberg, Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society: Series B (Methodological) 57, 289-300, doi:https://doi.org/10.1111/j.2517-6161.1995.tb02031.x (1995).

Competing interest reported. IN is serving on scientific advisory boards for Biogen Japan and Novartis Pharma and is receiving honouraria for speaking engagements with Biogen Japan, Mitsubishi Tanabe Pharma, Novartis Pharma, Takeda Pharmaceutical, and Eisai. JF have no conflicts of interest to disclose.

Download PDF

Version 1

posted

You are reading this latest preprint version

Early-stage volume losses in the corpus callosum and thalamus predict the progression of brain atrophy in patients with multiple sclerosis

Status:

Version 1

Abstract

Figures

Introduction

Results

Patient clinical profiles

Cluster analysis identified three clusters based on global and regional brain volume data

Comparisons of clinical and radiological findings among the three clusters at baseline

Comparisons of clinical and radiological findings among the three clusters at follow-up

Comparison of baseline data between transition and nontransition groups

Predictor Screening

Discussion

Methods

Patients

MRI acquisition

MRI postprocessing to measure global and regional brain volumes

Cluster analysis of regional and global brain volumes

Statistical analysis

Abbreviations

Declarations

References

Additional Declarations

Status:

Version 1