Study protocol. A screening phase (weeks −8, −4, and 0) was followed by the treatment period. Thirty subjects each received 4 doses of rituximab (375 mg/m2) 1 week apart. Magnetic resonance imaging (MRI) of the brain with and without gadolinium and the Multiple Sclerosis Functional Composite (MSFC) were assessed serially in all subjects. At least 3 “practice” MSFC tests were performed before the baseline MSFC test at week 1 (just before the first infusion). Twenty-six subjects underwent cerebrospinal fluid sampling 1 week before and 24 to 30 weeks following rituximab treatment.
Effects of rituximab therapy on B- and T-cell levels in the cerebrospinal fluid (CSF) and blood. The B and T cells in the CSF and blood were counted before and after rituximab treatment by flow cytometry. A, Levels of B cells in the CSF decreased after treatment (P < .0001 by Wilcoxon matched pairs test). B, CD19+ B cells were eliminated from the blood by 8 weeks in all subjects, and most remained depleted of B cells at weeks 24 to 30. C, Numbers of CD3+ T cells in the CSF decreased in 21 of the 26 subjects who underwent the lumbar puncture after rituximab (P = .0001, Wilcoxon matched pairs test). D, The number of T cells in the blood also showed a small but significant reduction after rituximab treatment (P = .0008 by Wilcoxon matched pairs test).
Effect of rituximab therapy on CXCL13 and CCL19 levels in the cerebrospinal fluid (CSF) and blood. A, Levels of CXCL13 in the CSF were measured by enzyme-linked immunosorbent assay (ELISA) before and after rituximab treatment in 23 subjects, and the levels decreased in all but 1 case (P = .002). B, Serum levels of CXCL13 (n = 25 pairs) also declined significantly (P < .0001), with a decrease in 21 subjects, an increase in 2, and no change in 2. C, Levels of CCL19 were measured by ELISA in 26 sets of paired CSF and, D, in 13 sets of paired serum samples obtained before and after rituximab treatment. Levels of CCL19 in the CSF and serum significantly declined after rituximab treatment (P = .03 and P = .008, respectively).
The degree of cerebrospinal fluid (CSF) T-cell level reduction correlated with the degree of CSF CXCL13 level decrease. The percentage of changes of CD3+ T-cell levels and CXCL13 levels in CSF were determined for each patient before vs after rituximab treatment (n = 23) and were found to be correlated (r = 0.45; P = .03, Spearman rank correlation). The best-fit line is shown.
Piccio L, Naismith RT, Trinkaus K, Klein RS, Parks BJ, Lyons JA, Cross AH. Changes in B- and T-Lymphocyte and Chemokine Levels With Rituximab Treatment in Multiple Sclerosis. Arch Neurol. 2010;67(6):707-714. doi:10.1001/archneurol.2010.99
B cells are implicated in the pathogenesis of multiple sclerosis. A beneficial effect of B-cell depletion using rituximab has been shown, but the complete mechanism of action for this drug is unclear.
To determine the relationship between T and B cells and changes in cerebrospinal fluid (CSF) chemokine levels with rituximab, a monoclonal antibody that targets CD20.
Phase 2 trial of rituximab as an add-on therapy.
The John L. Trotter Multiple Sclerosis Center, Washington University.
Participants and Intervention
Thirty subjects who had relapsing-remitting multiple sclerosis with clinical and magnetic resonance imaging activity despite treatment with an immunomodulatory drug received 4 weekly doses of rituximab (375 mg/m2).
Main Outcome Measures
Lumbar puncture was performed before and after rituximab infusions in 26 subjects. Levels of B and T lymphocytes in the CSF were enumerated by flow cytometry, and chemoattractant levels were measured by enzyme-linked immunosorbent assay.
After rituximab administration, CSF B-cell levels were decreased or undetectable in all subjects, and CSF T-cell levels were reduced in 21 subjects (81%). The mean reduction in CSF cellularity was 95% for B cells and 50% for T cells. After rituximab infusion, CSF CXCL13 and CCL19 levels decreased (P = .002 and P = .03, respectively). The proportional decline in CSF T-cell levels correlated with the proportional decrease in CXCL13 levels (r = 0.45; P = .03), suggesting a possible relationship. The CSF IgG index, IgG concentration, and oligoclonal band number were unchanged following treatment.
In subjects with multiple sclerosis, B cells are critical for T-cell trafficking into the central nervous system and may alter the process by influencing chemokine production within the central nervous system.
Multiple sclerosis (MS) is a central nervous system (CNS) disorder that affects 2.5 million people worldwide. The pathogenesis of MS is still not fully understood. Multiple sclerosis had been thought to be mediated primarily by autoreactive T cells, but recent findings have indicated a critical role for B cells.1,2 Abnormal B-cell activity is a prominent feature in MS, and B cells and numerous plasma cells can be observed in active3,4 and chronic5 MS lesions. Intrathecal synthesis of immunoglobulins is a diagnostic feature of the disease.6 In addition, B cells could play a role through antibody-independent mechanisms that include antigen presentation, costimulation, and cytokine and chemokine production.7
A recent phase 2 trial of B-cell depletion using rituximab, a monoclonal antibody targeting CD20, demonstrated decreased clinical and imaging activity in subjects with relapsing-remitting MS (RRMS) when they received it as monotherapy.8 In the present phase 2 study, rituximab was used as an add-on therapy to deplete B cells in subjects with RRMS who had breakthrough inflammatory activity despite standard immunomodulatory drug therapy. In an earlier report, we demonstrated a reduction in T- and B-cell levels in the cerebrospinal fluid (CSF) in the first 15 subjects treated.9 The completed clinical trial included 26 subjects who had undergone pretreatment and posttreatment lumbar punctures (LPs). A significant reduction of CSF T-cell levels was again observed. This was an unexpected finding because rituximab specifically targets only B cells. One possible reason for the T-cell level decline is through a reduced production of chemoattractant factors in the CNS, directly or indirectly by B cells. Thus, levels of 17 candidate chemokines and chemoattractant factors, selected because they are produced by B cells or known to be elevated in the CSF in MS, were compared in CSF before and after rituximab treatment. Levels of CXCL13 and CCL19, 2 chemokines that are involved in the organization of lymphoid follicles,10- 12 declined significantly following treatment.
This phase 2 trial was designed to study rituximab as add-on therapy in patients with RRMS who had ongoing disease activity despite therapy with an immunomodulatory medication approved by the US Food and Drug Administration. The study was approved by the Washington University Human Research Protection Office. All subjects provided informed consent. Inclusion criteria included RRMS with an expanded disability status score of 6.5 or less, exacerbation within 18 months despite receiving at least 6 months of immunomodulatory drug therapy, and at least 1 gadolinium-enhancing lesion on any of 3 pretreatment monthly magnetic resonance images of the brain. Prior treatment with an immunosuppressive agent, other than periodic corticosteroid therapy, was exclusionary. Patients had to be free of corticosteroid therapy for 30 days before the initial screening. Rituximab was administered at 375 mg/m2 per week for 4 doses. Subjects continued their immunomodulatory drug therapy (Table 1). The Multiple Sclerosis Functional Composite (MSFC)13 was administered at each visit. Three or more “practice” MSFC tests were obtained before the baseline MSFC examination. Twenty-six subjects underwent CSF and blood sampling 1 week before and 24 or more weeks after the initial dose of rituximab (Figure 1). The CSF samples were assessed for IgG concentration, number of oligoclonal bands, IgG synthesis rate,14 and IgG index (reference normal cutoff, <0.68) by the hospital laboratory.
Rituximab is a chimeric murine/human IgG1 κ monoclonal antibody that targets CD20, a transmembrane phosphoprotein expressed only by pre-B and mature B cells.15 Rituximab lyses B cells via complement-dependent16 and antibody-dependent15,17 cellular cytotoxicity.
Cerebrospinal fluid cells were examined by flow cytometry (FACSCalibur flow cytometer; Becton Dickinson Biosciences, Downers Grove, Illinois) within 5 hours of LP. After dividing the CSF into aliquots for immunoglobulin analyses, the remainder was centrifuged, and supernatants were frozen at −80°C for chemokine analyses. The T and B cells were identified with peridinin chlorophyll protein–labeled antihuman CD3 and with allophycocyanin-labeled antihuman CD19 (both from BD Pharmingen, San Jose, California), respectively. The CSF cells were also stained for CD80, CD86, and CD138 and, in some cases, for CD25, CD38, CD27, or CXCR5 (BD Pharmingen). Isotype-matched antibodies served as controls. Data were analyzed with commercially available software (CellQuest; Becton-Dickinson, Franklin Lakes, New Jersey). For peripheral blood, white blood cells were isolated using a density gradient (Ficoll-Paque; Amersham Biosciences, Piscataway, New Jersey) and stained as described for CSF cells.
Aliquots of CSF were assayed in duplicate (kits from R&D Systems [Minneapolis, Minnesota] except C3a and C5a, from Becton-Dickinson). Intra-assay coefficients of variation ranged from 0.1% for CXCL13 to 4.4% for CXCL12 enzyme-linked immunosorbent assays.
Nonparametric Wilcoxon matched pairs tests were used to compare the numbers of B and T cells, proportion of CSF B cells expressing phenotypic markers, and levels of CXCL13, CCL19, and other chemokines at baseline and after treatment. Spearman rank correlations were used to examine the relationship between CSF T- and B-cell numbers and chemokine levels, to address relationships between proportional changes in T-cell numbers with changes in CXCL13 and CCL19 levels, and to test for associations of changes in B- and T-cell counts in CSF with changes in MSFC scores. P values were adjusted for multiple comparisons using the stepdown Bonferroni approach.
Thirty subjects (22 women and 8 men) received 4 doses of rituximab (Table 1 and Figure 1). Twenty-six of these underwent LP before and after treatment. The posttreatment LP was 24 to 30 weeks after the first rituximab infusion, except in 3 subjects in whom it was delayed because of scheduling issues (33, 35, or 38 weeks).
Nineteen of the 26 subjects undergoing LP had undetectable B-cell levels in the blood at the time of the second LP; the other 7 had B cells constituting 1% to 11% of circulating mononuclear cell levels. The highest percentages were in subjects who delayed posttreatment LP. The CSF B-cell levels decreased after treatment in 20 subjects (Figure 2A; P < .0001 by Wilcoxon matched pairs test). In the remaining 6 subjects, B cells were undetectable in CSF before or after rituximab infusion. In no case did B-cell levels increase in CSF after treatment.
The CSF T-cell levels declined in 21 of the 26 subjects after treatment (P = .0001, Wilcoxon matched pairs test) (Figure 2C) and remained unchanged in 1 subject. The CSF T-cell levels increased in 4 subjects after treatment, 2 of whom were among the 6 individuals with no detectable CSF B-cell levels before treatment. Overall, CSF T-cell levels declined by more than 50% compared with pretreatment measurements. The CD3+ T-cell levels in the blood also declined by a mean of 12% after rituximab treatment (P = .0008; Figure 2D), a finding that has not previously been reported. In individual subjects, there was no correlation between T-cell numbers in the blood and those in CSF before or after rituximab treatment.
No differences were noted in CSF B- and T-cell numbers before or after treatment between subjects taking interferon beta vs glatiramer acetate. Baseline CSF T- and B-cell numbers did not correlate with pretreatment MSFC scores or any of the individual component test scores. No correlations between the percentage of change in T-cell levels and the percentage of change in MSFC scores or any of the individual component test scores were found.
Cerebrospinal fluid B-cell expression of CD80, CD86, CD25, CD38, and CD138 was examined by flow cytometry (Table 2). Although the number of CSF B cells was greatly decreased after treatment, the proportion of CSF B cells that also expressed the costimulatory molecules CD80 and CD86 was significantly increased (P = .01 for both). No other significant differences were seen in CSF B cells following treatment. No significant changes in expression of CD25, CD27, CD38, CXCR5, and CD45RO were observed in CSF T cells.
Immunoglobulin levels of the CSF, including IgG index, IgG concentration, IgG synthesis rate, and oligoclonal bands, did not change significantly after treatment.9
The decline in CSF T-cell levels was unexpected because rituximab targets CD20, which is restricted to B cells. We hypothesized that B cells in the CNS might produce or influence production of T-cell chemoattractants affecting T-cell trafficking into the CNS. Therefore, we measured levels of 17 candidate chemokines and chemoattractant factors in CSF. Nine of these—CCL2, CCL4, CCL19, CXCL10, CXCL12, CXCL13, CXCL16, interleukin 16, and C3a—were present at sufficient levels in CSF for accurate measurement (Table 3). The other 8 factors—CCL3, CCL5, CCL21, CCL22, CXCL9, CXCL11, lymphotoxin α, and C5a—were undetectable or detected at levels in the CSF that were too low to be reliable. Levels of 2 of the 9 detectable chemokines declined significantly in the CSF following rituximab therapy: CXCL13 (P = .002) and CCL19 (P = .03) (Table 3). Levels of the other 7 chemokines that could be measured did not change significantly when CSF samples obtained before vs after treatment were compared.
Of 23 paired pretreatment and posttreatment samples tested, CSF CXCL13 levels declined after treatment in all but 1 pair (P = .002; Figure 3A). That subject was notable for showing no decline in CSF T-cell numbers after treatment as well. Levels of CXCL13 were assayed in the serum in 25 paired samples before and after rituximab treatment and declined significantly overall in the blood (P < .0001), with a decrease in 21 subjects, an increase in 2, and no change in 2 (Figure 3B). No correlations between changes in CXCL13 serum and CSF levels were found for paired samples from individual subjects (r = 0.32; P = .14), and no difference in CXCL13 levels was noted before or after treatment in subjects taking interferon beta vs glatiramer acetate.
The level of CCL19 was tested in 26 sets of paired CSF and 13 sets of paired serum samples obtained before and after rituximab treatment. Levels of CCL19 in the CSF and serum each declined overall after rituximab treatment (P = .03 and P = .008, respectively). Levels of CCL19 in the CSF after treatment with rituximab declined by at least 10% in 17 subjects, increased by at least 10% in 4 subjects, and were unchanged in 5 (Figure 3C). Serum CCL19 levels declined after treatment in 9 subjects, increased in 1, and remained stable in 3 (Figure 3D). Changes in CCL19 serum and CSF levels were not correlated in individual subjects (r = 0.32; P = .36). When subjects taking interferon beta vs glatiramer acetate were compared, no difference in CCL19 levels was noted before or after treatment.
We reasoned that, if the decline in CXCL13 or CCL19 was related to the decline in T-cell levels, then the proportional decreases would be correlated. Thus, the percentage of change in CD3+ T-cell numbers in CSF was determined for each subject, as was the percentage of change in CXCL13 and CCL19 levels. The proportional change in CSF T-cell levels was mildly correlated with the proportional change in CXCL13 levels, suggesting that this chemokine was related to the decline in T-cell numbers (r = 0.45; P = .03) (Figure 4). There was no significant correlation between percentages of change in CSF T-cell levels and CSF CCL19 levels (r = 0.30; P = .17). There was a negative correlation between the change in T-cell numbers and the change in CSF monocyte chemotactic protein 1 levels (r = −0.46; P = .03). No relationships between proportional change in T-cell numbers and changes in the other chemokines were identified.
Because initial B-cell numbers in CSF before treatment were small, the percentage of change in B-cell levels could not be precisely determined, and no correlations of percentage of change in CSF B-cell levels with changes in chemokine levels or T-cell numbers were considered interpretable.
Until recently, evidence linking B cells and their products to MS pathogenesis has mainly consisted of observational studies. The availability of rituximab to specifically deplete B cells has provided a means to examine the pathogenic role of B cells in MS. Two studies published in 2008 indicated that B-cell depletion with rituximab resulted in diminished clinical and magnetic resonance imaging activity in patients with RRMS.8,18 In the present trial, 30 subjects underwent B-cell depletion with rituximab while they continued immunomodulatory drug therapy. B cells in the CSF and peripheral blood were eliminated. Unexpectedly, a decline in CSF and blood T-cell levels also occurred. The decline in T-cell as well as B-cell levels might be a mechanism through which rituximab benefits patients with RRMS.
Previously, B cells have been implicated in MS through their ability to produce antibodies. Correlations of high levels of CSF antibodies with poor outcomes in MS have been shown.19 In the present study, no changes in CSF IgG levels or oligoclonal band number were noted after B-cell depletion, suggesting that the effectiveness of rituximab is independent of CSF IgG levels. Thus, the present study supports not only the notion that B cells are critical to development of inflammatory lesions in relapsing MS but also that the role of B cells is not limited to antibody production. Indeed, B cells have several other functions that may be important, including antigen presentation20 and cytokine and chemokine production.21
After observing that T-cell levels declined in the CSF of subjects treated with rituximab, we measured CSF levels of several chemokines and chemoattractant factors that have been implicated in T-cell trafficking in MS. Levels of CXCL13, a chemoattractant factor for B cells and activated T cells that is involved in the organization of lymphoid follicles,10- 12 declined significantly following treatment. Of interest, there was a correlation between the degrees of reduction in CSF levels of T cells and CXCL13, suggesting an association between these 2 changes. Depletion of B cells alone would not explain the decreased levels because B cells are not known to produce CXCL13. Follicular dendritic cells (DCs) are an established source of CXCL13 within human lymphoid tissues.22 Published observations indicate that the cellular source of CXCL13 in the CNS is likely to be DCs or monocytes/macrophages. In immunohistochemical studies of autopsied active MS lesions, CXCL13 was observed within inflammatory cells having macrophage morphologic features.23 Furthermore, monocytes/macrophages are the source of CXCL13 in experimental autoimmune encephalomyelitis.24 Mice lacking CXCL13 develop less severe experimental autoimmune encephalomyelitis with less CNS inflammation, indicating an important role for this chemokine in the pathogenesis of the disease.24 T cells themselves might be a source of CXCL13. Recent human studies have detected the expression of CXCL13 by antigen-experienced helper T cells within synovial follicles and by TH17 clones.25,26 Thus, a reduction in trafficking of T cells to the CNS might amplify the decline in CSF CXCL13 levels.
The receptor for CXCL13 is CXCR5, expressed by most B cells, subsets of CD4+ T cells, and some DCs. Studies in subjects with MS23 demonstrated that about 20% to 30% of blood and CSF CD4+ T cells were CXCR5+, whereas virtually all B cells in the blood and most in the CSF were CXCR5+. Thus, the decline in CXCL13 levels would be expected to affect CNS trafficking of a significant proportion of T cells and most B cells.
Levels of CCL19 were also decreased in the CSF and serum of patients with MS after treatment with rituximab. Macrophages and DCs produce CCL19 in T-cell zones of secondary lymphoid organs, where it is critically involved in the migration of lymphocytes and mature DCs.27 CCL19 is constitutively expressed in the CNS,28- 30 and CCL19 may be critical for CNS immune surveillance. Levels of CCL19 in the CSF were increased in subjects with MS and those with other CNS inflammatory diseases compared with control subjects,29,30 and CCL19 RNA was elevated in active and inactive MS lesions.29
Expression of CXCL13 and CCL19 in MS CNS would be enhanced by the B-cell–derived factor lymphotoxin α,10,31 which is a key inducer of CCL1932 and CXCL13.33 In other autoimmune diseases, CXCL13 promotes B-cell expression of lymphotoxin α, which promotes follicular DC development and further expression of CXCL13, in a positive feedback loop.10 We speculate that greatly decreased levels of lymphotoxin α would result from B-cell depletion in the CNS, leading to decreased levels of CXCL13 and CCL19. However, we were unable to reliably measure lymphotoxin α in the CSF, so this hypothesis remains to be tested.
Reductions in CCL19 and CXCL13 levels following therapy with rituximab were notable because of recent renewed interest in lymph node–like structures in the MS CNS. Both CXCL13 and CCL19 have been implicated in the development of ectopic lymphoid follicles.34 Ectopic lymphoid-like structures have been seen in some MS plaques,35 and the endothelium in MS lesions sometimes resembles that of secondary lymphoid organs. Follicle-like structures, described in the meninges of some secondary progressive MS cases,34,36 have been associated with an aggressive clinical course.36 Because its expression has been detected in intrameningeal follicles in MS34,36 and in active MS lesions, CXCL13 is of special importance.23 Together, these findings suggest a CNS microenvironment in MS that would attract B cells and promote their expansion, maturation, and production of immunoglobulins and cytokines/chemokines and recruit T cells.
Although the present study indicates chemokine modulation as an important function of B cells in MS, this is unlikely to be the only role of B cells in MS pathogenesis. In addition, B cells can function as potent antigen-presenting cells to T cells, especially in situations of low antigen density.37 Loss of B cells as essential antigen-presenting cells in the periphery would result in decreased activation of T cells and decreased trafficking to the CNS. However, if this was the only mechanism of B-cell depletion leading to diminished immune cell trafficking, we would have expected a generalized decrease in inflammatory chemokines in CSF, not selectively CXCL13 and CCL19.
These studies provide further support for a critical role of B cells in MS and yield additional insight into the nature of that role. Depletion of B cells by rituximab reduced not only CSF B-cell levels, but also T-cell levels. This effect of B cells may be indirect, involving the reduced CNS production of chemoattractant factors such as CXCL13 by non-B cells. In addition, B cells may play a role as antigen-presenting cells to autoreactive T cells. In this current era of increasing numbers of monoclonal antibody therapies, the present study underscores that, even when targeting a single molecule restricted to a single cell type, there may be important downstream effects that may relate to its mechanism of action.
Correspondence: Anne H. Cross, MD, Department of Neurology, Washington University School of Medicine, Campus Box 8111, 660 S Euclid Ave, St Louis, MO 63110 (firstname.lastname@example.org).
Accepted for Publication: October 25, 2009.
Author Contributions: Dr Cross had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Piccio, Naismith, Lyons, and Cross. Acquisition of data: Piccio, Naismith, Parks, and Cross. Analysis and interpretation of data: Piccio, Trinkaus, Klein, Lyons, and Cross. Drafting of the manuscript: Piccio and Cross. Critical revision of the manuscript for important intellectual content: Piccio, Naismith, Trinkaus, Klein, Parks, Lyons, and Cross. Statistical analysis: Trinkaus. Obtained funding: Cross. Administrative, technical, and material support: Piccio, Trinkaus, Klein, Parks, and Lyons. Study supervision: Piccio, Naismith, Parks, and Cross.
Financial Disclosure: Dr Cross serves on the Research Programs Advisory Committee and the National Clinical Advisory Board of the National MS Society and the scientific advisory board for Eli Lilly and Company and BioMS; has received speaker honoraria for non–industry-sponsored activities; serves on the speakers' bureaus of Bayer HealthCare Pharmaceuticals (formerly Berlex Inc), Genentech, Inc, Biogen Idec, and Teva Neuroscience; has received consulting fees from Hoffman-La Roche; and receives research support from the National Institutes of Health (NIH) (NINDS PO1 NS059560- 01 [overall principal investigator (PI); PI of project 3 and core A], NINDS UO1 NS45719-01A1 [coinvestigator]; RO1 NS047592 [coinvestigator]; and NINDS/National MS Society RO1 NS 051591/NMSS RG 3915-A-15 [PI]) and from the National MS Society. She is on the editorial boards of Brain Pathology and Journal of Neuroimmunology and received an honorarium from the American Academy of Neurology (AAN) for editing and cowriting 2 chapters in CONTINUUM (Philadelphia, PA: Lippincott Williams & Wilkins; 2007). Dr Cross is also the Washington University site PI for clinical trials sponsored by Acorda Therapeutics and Sanofi-Aventis. Dr Naismith has served on speakers' bureaus and as a consultant for Bayer Healthcare, Biogen Idec, Elan Pharmaceuticals, and Teva Neurosciences; receives research support from Acorda Therapeutics (site PI) and the NIH (K23NS052430-01A1[PI] and K12RR02324902 [PI]); and received an honorarium from the AAN for editing and writing 1 chapter in CONTINUUM. Dr Klein serves on the Research Committee of the National MS Society and receives research support through the Washington University/Pfizer Biomedical Program, the National MS Society (RG3982), the DANA Foundation, and the NIH (NINDS PO1 NS059560- 01 [PI of project 2 and core B]). Dr Parks has served as a consultant and/or on speakers' bureaus for Bayer Healthcare, Biogen Idec, EMD Serono, and Teva Neuroscience.
Funding/Support: This study was supported by grant RG 3292 from the National MS Society USA; grants K24RR017100 and 5UL1 RR024992 from the NIH; the Barnes-Jewish Hospital Foundation; Genentech, Inc; and Biogen Idec. Dr Cross was supported in part by the Manny and Rosalyn Rosenthal–Dr John L. Trotter Chair in Neuroimmunology. Dr Piccio was supported by a postdoctoral fellowship from the National MS Society (FG 1665-A-1) and in early studies by Fondazione Italiana Sclerosi Multipla (FISM; 2004/B/4). Dr Naismith was supported by grants K23NS052430-01A1 and K12RR02324902.
Additional Information: The article is dedicated to the memory of this MS center's founder, John L. Trotter, MD (1943-2001).
Additional Contributions: Robert Mikesell, BS, Michael Ramsbottom, BS, and Neville Rapp, PhD, provided excellent technical assistance; Joanne Lauber, RN, Cathie Martinez, LPN, Monica Fairbairn, BS, and Nhial Tutlam, MPH, study subject coordination and MSFC testing; and John H. Russell, PhD, and Sheng-Kwei (Victor) Song, PhD, helpful discussions. We thank our patients for their participation.