In this observational case-control study, 40 treatment-naïve, newly diagnosed patients with RRMS (29 females/11 males; mean 36 years, range 21-62 years) and 40 patients treated with alemtuzumab (31 females/9 males; mean 39 years, range 22-56 years) at the Danish Multiple Sclerosis Center at Copenhagen University Hospital - Rigshospitalet were included; there were no significant differences in age or sex distribution between groups. The alemtuzumab-treated patients were all included in the recently published national study of Danish patients treated with alemtuzumab (11). The alemtuzumab-treated patients had a mean disease duration of 12 years (range 4-23 years), had a mean EDSS score of 4 (range 2-6.5), and had a mean number of 1.7 relapses (range 0-7) the year prior to the initiation of alemtuzumab therapy. They had received a mean of four previous therapies (range 1-7), and the most recent treatment was fingolimod in 19 patients, natalizumab in 15 and dimethyl fumarate or daclizumab in six. All patients fulfilled the 2017 McDonald criteria for RRMS (12). The alemtuzumab treatment regimen included a first course where 12 mg was infused on five consecutive days. Twelve months later, a second course of 12 mg daily was infused on three consecutive days. Blood samples were obtained from alemtuzumab-treated patients one year later to measure the repopulation of lymphocytes, i.e., two years after initiation of treatment.

All participants gave informed, written consent to participation. The study was approved by the regional scientific ethics committee (protocol number H-16047666).

The alemtuzumab-treated patients were seen for clinical controls at six-monthly intervals during the first three years after initiation of treatment and underwent magnetic resonance imaging (MRI) scans of the brain before treatment and annually after the initiation of treatment. The occurrence of relapses (defined according to the 2017 McDonald criteria) and secondary autoimmune diseases was based on chart review blinded to the results of the immunological studies. MRI activity was defined as new or enlarging T2 lesions compared to a rebaseline scan obtained one year after the initiation of treatment. No evidence of disease activity (NEDA) was defined as no relapses and no new or enlarging MRI lesions. Confirmed worsening of disability (CDW) was defined as a 1 point increase in Expanded Disability Status Scale (EDSS) score for patients with a baseline EDSS score from 1 to 5 and 0.5 for patients with a baseline EDSS score of 5.5 or more, confirmed after at least six months of follow-up (none of the patients had a baseline EDSS score of 0).

Venous blood was collected, and peripheral blood mononuclear cells (PBMCs) isolated by density gradient centrifugation using Lymphoprep (Axis-Shield, Oslo, Norway) and washed twice in cold PBS/2 mM EDTA within 1 hour of sampling. PBMC were then incubated with FcR blocking reagent (Miltenyi Biotec, Bergisch Gladbach, Germany) to prevent nonspecific Ab binding, and thereafter stained in PBS/2% FBS/0.02% NaN₃ with a combination of fluorochrome-conjugated antibodies targeting B cells, T cells and NK cells (conjugate; clone): CD3 (APC, APC/Cy7; UCHT1), CD4 (APC/Cy7, PerCP/Cy5.5; OKT4), CD8 (BV605; RPA-T8), CD19 (PerCP/Cy5.5; HIB19), CD25 (PE; M-A251), CD27 (FITC; 323), CD38 (BV421; HIT2), CD45RA (FITC; HI100), CD57 (PB; HNK-1), CD127 (APC; A019D5), CD183 (CXCR3, PE/Cy7; G025H7), CD185 (CXCR5, AF488; J252D4), CD196 (CCR6, BV421; G034E3), CD197 (CCR7, PE; G043H7), CD279 (PD-1, BV605; EH12.2H7) all from BioLegend (CA, USA) and CD56 (PE/Cy7; CMSSB) from eBioscience (ThermoFisher, MA, USA). CD20 on T cells was stained using either CD20-PE/Cy7 or BV421 (2H7, BioLegend) resulting in variability of the CD20-signal. An equal number of untreated and treated patients were stained with each of the two fluorochrome-conjugated antibodies; and depicted in Figure 1 where diamonds represent CD20-PE/Cy7 and circles CD20-BV421. Isotype matched controls were used to correct for nonspecific Ab binding and spectral overlap, where appropriate. TruCount staining of whole blood to measure absolute cell count was performed using BD Multitest 6-color TBNK Reagent according to manufacturer (BD Biosciences, San Jose, CA, USA). Data were acquired on a FACS Canto II flow cytometer (BD Biosciences) and data analyses performed using the software FlowJo (TreeStar, Ashland, OR, USA). Data analyses were performed blinded.

For analysis of sex differences between groups a Chi-square test was performed, and for analysis of age differences between groups a Mann-Whitney U test was applied. To compare cell frequencies and absolute number of cells between treatment naive and alemtuzumab-treated patients with RRMS a Mann-Whitney U test was performed. To analyze a possible association between immune cells and development of secondary autoimmunity, alemtuzumab-treated patients were divided into a group that developed autoimmune thyroid disease and one that did not and cell frequency, count or ratios of interest assigned to each group; and a Mann-Whitney U test applied. A similar analysis was performed investigating a possible association between immune cells and disease activity. As CD20⁺ T cells were analyzed using two different monoclonal antibodies, CD20⁺ T cells were analyzed in general linear models with the antibody used for analysis as a cofactor. Correlations were assessed by Spearman rank correlation analysis. A significance level of p < 0.005 was considered statistically significant for all analyses. For frequencies percentages are given with 95% confidence intervals (CI).

For this manuscript the STROBE reporting guidelines for observational studies was used (13).

Data are available in anonymized form and can be shared by request from any qualified investigator. Sharing requires approval of a data transfer agreement in accordance with GDPR and Danish data protection regulation.

In the 40 patients treated with alemtuzumab, 16 patients (40%, 95% CI 25-57%) had relapse activity in year one and 21 patients (53%; 95% CI 36-68%) had relapse activity in year two or three after initiation of treatment. The annualized relapse rate during the three years of follow-up was 0.6, i.e., substantially lower than the pre-treatment annualized relapse rate of 1.7. This is consistent with the notion that Danish patients treated with alemtuzumab, including the patients included in the present study, are characterized by having high disease activity and most patients having failed several therapies before initiation of alemtuzumab therapy (11). Twelve out of the 40 patients (30%, 95% CI 17-47%) had new or enlarging T2 lesions on MRI at year two or year three compared to the rebaseline MRI at year one. Thirteen patients (33%, 95% CI 19-49%) had no evidence of disease activity (NEDA) in year two and three.

Six patients (15%, 95% CI 6-30%) had confirmed EDSS worsening at the end of the three years of follow-up. Three of these patients had NEDA during the entire follow-up and three had either relapses or MRI activity during follow-up. The worsening was sustained to the end of follow-up in all patients.

A total of nine out of 40 patients (23%; 95% CI 11-38%) developed autoimmune thyroid disease during the three years of follow-up; none of the patients developed other forms of secondary autoimmune disease during follow-up. Two patients had autoimmune thyroiditis and seven had Grave’s disease and were seropositive for TSH receptor antibodies.

After two years, lymphocyte counts were not fully recovered (p < 0.001, Figure 1A ), and the distribution of reconstituted lymphocyte populations was profoundly altered in patients treated with alemtuzumab ( Figure 1B ). This included a reduced frequency of CD4⁺ T cells (p < 0.0001, Figure 1C ) and consequently a decreased CD4:CD8 T cell ratio (p < 0.0001, Figure 1E ), a reduced frequency of CD20⁺CD4⁺ and CD20⁺CD8⁺ T cells (p < 0.0001, Figures 1F, G ), an increased frequency of B cells (p < 0.0001, Figure 1H ) and NK cells (p < 0.0001, Figure 1J ), including CD56^hi NK cells (p < 0.0001, Figure 1L ), compared to untreated patients. The lower frequency of CD20⁺ T cell subsets was observed regardless of which antibody was used for staining.

The lower absolute lymphocyte count in reconstituted alemtuzumab-treated patients was reflected in a lower count of CD4⁺ and CD8⁺ T cells (p < 0.0001, Figure 1D ), CD20⁺CD4⁺ and CD20⁺CD8⁺ T cells (p < 0.0001), and NK cells (p = 0.0002) compared to untreated patients. In contrast, we observed a normalized number of B cells ( Figure 1I ), and an increased number of CD56^hi NK cells (p < 0.0001, Figure 1M ). Gating examples of T, B and NK cells are shown in Figures 2 , 3 and Supplementary Figure 1 .

Despite a normalized number of B cells following two years of alemtuzumab treatment, the distribution of B cell subsets was profoundly changed ( Figure 2I ). The frequency of naïve (CD27^-CD38^-/+) and transitional (CD27^-CD38⁺⁺) CD19⁺ B cells was increased (p < 0.0001); and memory B cells (CD27⁺CD38^-/+) and plasmablasts (CD27⁺⁺CD38⁺⁺) decreased (p < 0.0001; p = 0.0007) compared to untreated patients, Figures 2E, G . A gating example is shown in Figures 2A–D .

The same pattern was observed for absolute B cell subset counts with an increased number of naïve (p = 0.0004) and transitional (p <0.0001) B cells, and a decreased number of memory B cells (p < 0.0001), Figure 2F . The absolute count of plasmablasts was slightly but nonsignificantly reduced (p = 0.0160), Figure 2H .

Analyzing the distribution of naïve and memory T cells following two years of alemtuzumab treatment showed a decreased frequency of central memory (CCR7⁺CD45RA^-) CD4⁺ and CD8⁺ T cells (p < 0.0001); with no significant changes in either percent naïve (CCR7⁺CD45RA⁺), effector memory (CCR7^-CD45RA^-) or terminally differentiated (CCR7^-CD45RA⁺) T cell populations, Figures 3D, F . A gating example is shown in Figures 3A–C .

Although only the percentage of central memory T cells was affected by alemtuzumab treatment, the reduced total counts of CD4⁺ and CD8⁺ T cells resulted in a lower absolute number of naïve, central memory and effector memory CD4⁺ T cells (p < 0.0001, Figure 3E ) and of naïve, central memory, effector memory and terminally differentiated CD8⁺ T cells (p < 0.0001, Figure 3G ).

Regulatory T (Treg) cells act to suppress immune responses to maintain self-tolerance. The percentage of CD4⁺CD127^-CD25⁺⁺ Treg cells after two years of reconstitution was comparable to the frequency in untreated patients ( Figure 4B ); but due to the lower number of repopulated CD4⁺ T cells the absolute number of Treg cells was decreased (p < 0.0001, Figure 4C ). A gating example is shown in Figure 4A .

Effector T (Teff) cells can be subdivided into functionally different subsets according to their expression of CXCR3 and CCR6. Following two years of alemtuzumab treatment there was an increased frequency of CD4⁺ CXCR3⁺CCR6^- (Th1-like; p < 0.0001) and CD8⁺ CXCR3⁺CCR6^- (Tc1-like; p < 0.0001) T cells compared to untreated patients, Figures 4D, F . Conversely, we found a reduced frequency of CD4⁺ CXCR3⁺CCR6⁺ (Th17.1-like; p < 0.0001), CD8⁺ CXCR3^-CCR6⁺ (Tc17-like; p < 0.0001) and CD8⁺ CXCR3⁺CCR6⁺ (Tc17.1-like; p < 0.0001) T cells, Figures 4D, F . As a result of the lower number of repopulated CD4⁺ and CD8⁺ T cells, all six CXCR3 CCR6 populations were significantly reduced in absolute numbers, Figures 4E, G .

A skewing towards a lower T effector:Treg cell ratio towards the tolerogenic Treg component, signifies a more immunosuppressive environment with potential for increased control of T effector cell activation. After two years of alemtuzumab treatment, we observed a significantly reduced ratio between Th17, Th17.1, Tc17 and Tc17.1-like cells and Treg cells (p < 0.0001), Figures 4H, I . No significant changes was observed in the Th1:Treg or Tc1:Treg cell ratios, despite an increased frequency of both Th1 and Tc1 cells. CD20⁺ T cells are effector T cells of a Th1/Tc1-like phenotype with a proposed proinflammatory role in RRMS (14–16). In this study, we observed a reduced ratio of both CD4⁺CD20⁺ T cells and CD8⁺CD20⁺ T cells and Treg cells (p < 0.0001) following two years of alemtuzumab therapy, implying an increased control of this subset along with the Th17/Th17.1/Tc17/Tc17.1-like subsets.

Conditions that cause prolonged activation of the immune system, including repopulation following immune-depletion therapy, may induce a state of T cell dysfunction known as T cell exhaustion (17). To analyze T cell exhaustion, we measured the loss of CD127 as a marker of exhaustion on effector T cells (17, 18). This showed a significant increase in the frequency of CD127^- CD4⁺ and CD8⁺ T cells (p < 0.0001, Figures 5D, E ). A gating example is shown in Figures 5A, B .

T cell exhaustion is also associated with expression of CD57 (19, 20). Analyzing CD57 expression on repopulated T cells from alemtuzumab treated patients showed a strong increase in CD57⁺ CD4⁺ central memory and effector memory T cells (p < 0.0001, Figure 5F ) and in CD57⁺ CD8⁺ effector memory T cells (p < 0.0001, Figure 5G ). Furthermore, a correlation analysis showed a direct correlation between CD57 upregulation and loss of CD127 on CD8⁺ T cells (p = 0.0006, r_s = 0.7). A gating example is shown in Figure 5C .

Follicular helper T (Tfh) cells, defined as CD4⁺CD127⁺CXCR5⁺PD1^+/- T cells (21), are important for B cell activation, antibody production and survival (22, 23). The Tfh subpopulation Tfh17 is of particular interest as it promotes activation of naïve B cells (24). CD25^int Tfh cells has previously been shown to represent CXCR3^-CCR6⁺ Tfh17 cells (21). In this study, we therefore defined Tfh17 cells as CD25^int Tfh cells, see Figure 6A . PD-1 is upregulated on the surface of Tfh cells upon cognate antigen activation (25). As T cells after two years of alemtuzumab therapy is characterized by extensive activation, PD-1 expression analysis was included. We observed that the frequency of Tfh cells was slowly normalizing ( Figure 6D ) after two years of treatment, in contrast to Tfh17 cells that were still suppressed (p < 0.0001, Figure 6H ). We also found, that most Tfh and Tfh17 cells expressed PD-1 in contrast to what was observed in untreated patients (p < 0.0001, Figures 6F, J ). Due to the lower absolute count of CD4⁺ T cells in treated patients all four Tfh-populations were reduced in numbers ( Figures 6E, G, I, K ).

The regulatory counterpart of Tfh cells, follicular regulatory T (Tfr) cells, defined as CD4⁺CD127^-CD25⁺⁺CXCR5⁺ T cells (21), have the potential for immune regulation of Tfh cells and hence of B cell activation including suppression of high-affinity antibody production (22, 23). Following two years of alemtuzumab treatment, Tfr cells were also strongly reduced both in percentage and absolute numbers (p < 0.0001, Figures 6B, C ).

The capacity of Tfr cells to control the activity of Tfh and hence of B cells can be evaluated by the Tfh : Tfr cell ratio. This showed a significant increase in the ratio between Tfh, PD-1⁺Tfh, Tfh17, PD-1⁺Tfh17 and Tfr cells, indicating a decreased control of all four Tfh-populations (p < 0.0001, p = 0.0006 for Tfh17 cells, Figures 6L–O ).

T cell exhaustion has been suggested as a marker of alemtuzumab-treated patients who develop secondary autoimmunity (26). Investigating an association between T cell exhaustion and development of thyroiditis showed no correlation between development of thyroid autoimmunity and either percent or absolute count of exhausted CD4⁺ or CD8⁺ T cells defined by loss of CD127 or gain of CD57. Furthermore, the extensive activation of Tfh and Tfh17 cells observed in this study as an upregulation of PD-1 was not associated with development of thyroiditis ( Table 1 ).

Thyroid autoimmunity is considered to be B cell driven and anti-thyroid antibody mediated (27); however, we did not find a correlation between development of thyroiditis and B cells or plasmablasts, Table 1 . The balance between Tfh and Tfr cells is important in maintaining immune tolerance primarily by controlling the activity of B cells (22). However, analyzing a possible association between the Tfh : Tfr cell ratio and development of thyroiditis also showed no correlation ( Table 1 ).

Proinflammatory CD20⁺ T cells are likely implicated in various autoimmune diseases (28). Analyzing a possible association between CD20⁺ T cells and development of thyroiditis showed no association between thyroid autoimmunity and either percentage or absolute counts of CD4⁺CD20⁺ or CD8⁺CD20⁺ T cells, nor of CD20⁺ T cell:Treg ratios ( Table 1 ).

In order to analyze the relationship between immune cell subsets and disease activity, we compared values in patients with NEDA in year two and three after initiation of treatment. In addition, we assessed any relationship with the number of relapses during the three years of treatment or the development of confirmed disability worsening (CDW).

A previous study indicated an association between IL-17 producing CD4⁺ T cells and patients experiencing a relapse following 18 months of alemtuzumab treatment (29). We therefore analyzed a possible association between NEDA, relapses and CDW and the prevalence of Th-subsets including IL-17 producing Th17 and Th17.1-like cell, and found no correlations ( Table 1 ). CD20⁺ T cells have also been associated with disease activity in patients with RRMS (14–16). We found some relationship between CD20⁺ T cells and disease activity in the entire material ( Table 1 ); however, we found no significant relationship in multivariable analysis with correction for the monoclonal antibody used for analysis (data not shown). Also, in the larger subgroup of 25 patients studied with the BV421-conjugated anti-CD20 antibody, we found no significant relationship between CD20⁺ T cells and NEDA or relapses. Neither did we find any relationship between any of the blood cell subsets and the total number relapses during the three years of treatment or the development of CDW (data not shown).