Repopulation of T, B, and NK cells following alemtuzumab treatment in relapsing-remitting multiple sclerosis

Currently approved disease modifying drugs for patients with relapsing-remitting multiple sclerosis (RRMS), an autoimmune demyelinating disease of the CNS, target cells of the immune system to achieve clinical efficacy [1, 2]. Most have been reported to act in variable fashion to suppress or block pro-inflammatory CD4+ T cell types and cytokines, such as IFN-γ-secreting Th1 cells and IL-17A-secreting Th17 CD4+ T cells, and to promote immune regulation via effects on classical foxP3+ Tregs, Th2 cells, and IL-10 secreting cells. MS drugs may also target CD8+ T cells [3, 4], natural killer (NK) cells [5], antigen presenting cells [6], and more recently, B cells [7‐9]. In addition, fingolimod [10], cladribine [11], glatiramer acetate [12], and others (reviewed by Longbrake and colleagues [13]) may differentially affect B cell subsets, especially memory B cells [14]. Characterization of persistent changes in immune cell types and functions in response to treatment has potential to advance our understanding of disease pathogenesis and to promote development of more effective and refined treatment strategies. Furthermore, long-term analyses of immune phenotypes that prevail or change over time could be crucial to the assessment of the risk of future disease activity.

Alemtuzumab (Lemtrada, Genzyme, Cambridge, MA) is a humanized monoclonal antibody against CD52, a glycoprotein expressed on the surface of most lymphoid, and to a lesser extent, myeloid cell types. It was approved in the USA for the treatment of RRMS in 2014, based on durable clinical efficacy, in spite of carrying a high risk for development of secondary autoimmunity [15‐19]. The standard treatment protocol consists of two annual courses of i.v. alemtuzumab (12 mg/day) for five consecutive days at baseline and three consecutive days 12 months later (M12), with provisions for additional 3-day courses as needed.

Alemtuzumab induces rapid and profound lymphopenia within days of infusion, accompanied by more subtle and transient effects on monocytes, NK cells, dendritic cells, and neutrophils [16, 20‐22]. Lymphopenia is followed by repopulation of B cells, CD8+ T cells, and CD4+ T cells in sequence over time, with persistent depletion of CD4+ T cells for 3 to 4 years [23‐25]. The temporal pattern of repopulation, accompanied by persistent redistribution of T cell subsets in favor of circulating memory and regulatory T cells and against pro-inflammatory cytokines and Th1 and Th17 subsets, are considered to be fundamental to mechanisms of action that lead to durable efficacy in the absence of ongoing treatment [24, 26‐30]. However, changes in B cells [22, 26, 31] and innate-like lymphoid cells [21], including NK cells, have also been observed, suggesting that efficacy and the development of secondary autoimmunity involve greater mechanistic complexity.

The goals of this multicenter cohort alemtuzumab treatment study were as follows: (1) to analyze the repopulation of immune cells over time, with a focus on the balance between regulatory T cell, B cell, and NK cell subsets and effector subsets in each compartment; (2) to assess Treg functional competence over time; (3) to identify changes in T cells that express chemokine receptors; (4) and to explore associations of immune measures with disease activity and the risk of secondary autoimmunity.

The data in this manuscript show that regulatory cell types that are enhanced and persist after alemtuzumab treatment include classical CD4+CD25^hiCD127^lo/negfoxP3+ Tregs, CD19+CD27-CD24^hiCD38^hi naïve “regulatory” B cells, and CD56^bright NK cells. All were present at higher percentages within their respective parent populations after treatment, compared with baseline. Although similar findings have been reported for each of these subsets, very few studies have assessed all of them in the same patient cohort in long-term studies. In addition, the data expand our knowledge of the effect of alemtuzumab on the balance between regulatory and putatively pathogenic effector cell types, in which the ratios of CD4+CD25+CD127+foxP3- Teff:Treg, CD19+CD20+CD27+ memory B cells:naïve “regulatory” B cells, and CD3+CD56+NKT:CD56^bright NK cells were significantly reduced, in favor of regulation.

The Treg observations in this study are consistent with several reports that CD4+CD25^hiCD127^lo/neg foxP3+ Tregs expand preferentially in the CD4+ T cell pool during the early stages of recovery, peaking at approximately 3–5 months and remaining elevated compared with baseline for approximately 2 years [23, 29, 30, 37]. Our findings expand these observations to include assessments of ratios between Teff:Treg, which were profoundly and persistently reduced over time. Teff:Treg ratios were similarly reduced when calculated on the basis of estimated Treg and Teff counts (data not shown).

The data provide both direct and indirect evidence of Treg competence. Direct evidence was provided in the form of significant rebound proliferation at M5 and M17 in PHA-activated PBMC depleted of Tregs, timepoints that follow cycles of alemtuzumab treatment at M0 and M12. Treg competence has been reported at 12 and 24 months post-treatment by De Mercanti and colleagues, who conducted enzyme-linked immunospot (ELISPOT) assays to detect rebound of myelin basic protein (MBP)-stimulated IFN-γ- and IL-17A-secreting cells in PBMC depleted of CD25+ cells [29]. In addition, Jones and colleagues employed classical Treg proliferation inhibition assays in alemtuzumab-treated patients to detect Treg competence at 36–48 months [37], but not earlier timepoints. Importantly, our data demonstrate that Tregs are not only functionally competent, but that they exhibit more robust competence at the early timepoints of M5 and M17.

Indirect evidence of Treg competence can be found in data showing that the majority of Tregs express CD39, an ectoenzyme associated with regulation of immune responses [60‐62]. Our findings confirm reports of CD39+CD25^hi foxP3+ Tregs in the repopulating CD4+ T cell pool in alemtuzumab-treated patients [30, 37, 63]. The importance of CD39 expression in Tregs is supported by reports that the presence of CD39 stabilizes and promotes functional capacity in Tregs [64‐67]. In addition, CD39+ Tregs have been reported to be reduced or functionally defective in MS patients [64, 68, 69], similar to original reports for classical Tregs [70, 71].

Although not directly assessed in this study, functionally competent regulatory B cells and CD56^bright NK cells may accompany the expansion of Tregs. Kim and colleagues reported that deficits in regulatory B cells, defined as CD19+CD24^hiCD38^hi or CD19+PD-L1^hi, are reversed up to 1 year following alemtuzumab treatment [22], and Grutzke and colleagues [10] have reported that CD19+CD27-CD38^hi B cells from MS patients secrete high levels of IL-10. In addition, Gross et al. reported that CD56^bright NK cells retain potentially regulatory cytolytic function at 6 months after alemtuzumab treatment [21]. Of special interest is that changes in phenotypically naïve “regulatory” B cells and CD56^bright NK cells were more robust, persisted for at least 2 years and exhibited less variability than those in Treg subsets, suggesting that they may play key roles in alemtuzumab’s durable mechanism of action.

Because regulatory populations in each compartment interact with each other to fine-tune regulation of immune responses, clinical efficacy in alemtuzumab-treated patients may also reflect the sum of a network of these interactions. For example, Tregs are capable of directly and indirectly suppressing B cells to control antibody production [72]. Regulatory B cell subsets not only inhibit Th1 and Th17 cells, but also convert CD4+ T cells into Tregs or IL-10 producing Tr1 cells via production of regulatory cytokines, including IL-10, IL-35, and/or transforming growth factor-beta (TGF-β) [73, 74]. In addition, subsets of NK cells promote Treg development [75‐77], and regulatory T, B, and NK subsets all interact with antigen presenting cells to promote a tolerogenic cascade [70, 74, 78, 79]. Furthermore, several subsets of regulatory CD8+ T cells [80] are capable of contributing to a regulatory environment.

Our study included assessments of changes in T cells that express chemokine receptors, initially as a surrogate measure of Th1 cells, which express high levels of CXCR3+ and CCR5+, and Th2 cells, which are more likely to express CCR3 and CCR5. We originally hypothesized that an anti-inflammatory immune environment following alemtuzumab treatment may consist of increases in CCR3+ and/or CCR5+ Th2 cells and decreases in CXCR3+ and/or CCR5+ cells. Our findings do not support the use of chemokine receptor expression as indicators of Th1 or Th2 bias in CD4+ T cells, especially since cytokine secretion patterns, or additional functional assessments, were not included. However, the data clearly indicate that CD4+ and CD8+ T cells are activated and show enhanced capacity for migration in alemtuzumab-treated patients. Future studies might benefit from more detailed functional and phenotypic characterization of chemokine receptor bearing T cells.

A recent study published by Wiendl and colleagues reported that no clear immunological signals of efficacy or secondary autoimmunity could be identified in alemtuzumab-treated patients from the pivotal CARE-MS trials up to 24 months after treatment [81]. Their study assessed lymphocyte measures based on cell counts and included multiple phenotypes, many similar to those included in our study, as well as clinical measures that included relapses, 6-month confirmed disability worsening (CDW) based on EDSS and MRI activity, consisting of new Gd+ T1 lesions in current MRIs or new T2 hyperintense lesions since the last MRI. Our findings are generally consistent with this report.

However, patients in our cohort with stable disease showed a trend for reduced CD4+CD25+CD127+foxP3- Teff cells, most evident at M5, and Teff:Treg ratios that remained reduced from M5 to M48. In addition, at month 11, prior to the second course of alemtuzumab infusions, the Teff:Treg ratio appeared to be reverting to baseline in patients with active disease. This also occurred, to a lesser extent, in patients with and without relapses or Gd+ and T2 lesions. Future studies with larger numbers of patients are needed to determine if lack of depletion or reduced depletion of CD4+CD25+CD127+foxP3- Teff occurs in patients who exhibit disease activity on alemtuzumab, and as such, might serve as a potential biomarker for a suboptimal treatment response. To our knowledge, Teff defined as CD4+CD25+CD127+foxP3- cells have not been monitored in clinical studies, but we considered them as possible representatives of a general population of activated T cells that either exhibit effector functions or contain specific subsets of effector T cells. Thus, it would be of interest to determine if CD4+CD25+CD127+foxP3- Teff express markers or functions characteristic of pro-inflammatory “effector” T cells, such as IL-17A and IFN-γ.

Since MS is clearly an immunopathogenic disease, it is difficult to understand why analyses using linear mixed effects models for repeated measures have failed to more clearly identify immune markers of clinical status, especially since a large number of patients and immune measures were included in the study published by Wiendl and colleagues. We are either not on the right track with selection of immune measures for analyses, or we need to develop a different strategy of thinking about the cell types that may be involved. In view of the dynamic, plastic and heterogeneic nature of immune responses in general, and in immune repopulation in response to lymphopenia, it seems reasonable to consider relative contributions of multiple cell types. The results of the exploratory random generalized estimating (GEE) Poisson analyses of the risk for Gd+ and T2 lesions associated with increases in immune measures support this proposition and provide preliminary guidance for identification of relevant immune measures in future studies in larger patient cohorts. Of note, we found that the risk for Gd+ lesions (expressed as a rate ratio > 1.0), increased significantly when the percentage of general T cell populations (including CD3+ and CD4+ CD4+ T cells, CD4+CD45RA+ total naïve T cells), CD3+CD56+ NKT cells, and CD19+CD20+CD27+ memory B cells increased by one standard deviation at any timepoint. By contrast, an increase in regulatory cell types, including naïve “regulatory” CD24^hiCD38^hi B cells, CD3-CD56^bright NK cell Tregs, and Treg function was associated with reduced risk for Gd+ lesions, or a protective effect (expressed as a rate ratio < 0.5). These findings support the possibility that potentially pathogenic “effector” subsets in the T cell, B cell, and NK cell compartments contribute to inflammatory lesion activity and that, as we have proposed, multiple regulatory T, B, and NK subsets and functions are capable of contributing to a protective effect.

In most patients, concentrations of IL-2 in supernatants from PHA-stimulated PBMC were dramatically reduced from M5 to M23, along with IFN-γ and IL-17A. Several investigators have assessed cytokine secretion patterns in serum and PBMC from alemtuzumab-treated patients [23, 29, 30, 37, 82], but not by CBA in supernatants from PHA-activated PBMC. Reductions in IL-2, IFN-γ, and IL-17A may reflect persistent depletion of total circulating CD4+ T cells, or, especially for IL-2, could be due to heightened susceptibility of repopulating T cells to activation-induced cell death following alemtuzumab treatment [83]. However, an increase in IL-2 concentrations was associated with increased risk for Gd+ lesions (rr = 1.33, p = 0.048) with possible implications for a less than adequate response to alemtuzumab. Interestingly, an increase in IL-10 concentrations yielded a rate ratio of 0.58 (p = 0.04) for Gd+ lesions, suggesting a possible protective effect.

CD3+CD8+CXCR3+ cells were the only cell type that showed a significant association with T2 lesions, which are generally a reflection of longer-term demyelination and axonal damage, and not active or acute inflammation. CD8+ T cells have long been known to outnumber CD4+ T cells in the CNS of MS patients and there is clear evidence of associations of CD8+ T cell subsets with demyelination and axonal damage [84‐87]. Although total CD3+CD8+ T cells did not show a significant association with T2 lesions in the current study, the data suggest different mechanisms of immunopathogenesis for Gd+ and T2 lesions, and thus, support more thorough investigation of CD8+ T cells in the future.

Although we interpret the compelling preliminary findings from the exploratory random GEE Poisson analyses cautiously, they have identified several putative early immunologic biomarkers with potential to identify patients at risk for disease activity. It would be of great interest to apply this type of analyses to the substantially larger CARE-MS datasets, not only to confirm and validate the relative risk of lesion activity with pathogenic and regulatory lymphocyte subsets, but to determine if immunological signals of a treatment effect, or secondary autoimmunity can be identified.

At least 40% of patients treated with alemtuzumab develop secondary autoimmunity following alemtuzumab treatment and the incidence peaks approximately 3 years after the first treatment course [19]. This is not typically seen with other MS therapies that cause lymphopenia or specifically target B cells, with perhaps an exception for daclizumab, which targets NK cells [88]. We did not observe differences in T cell subsets or Treg function in patients with and without secondary autoimmunity, in agreement with others [37, 81]. It has been suggested that hyper-repopulation of B cells, especially naïve B cell subsets, in the absence of T cell regulation predisposes patients to secondary autoimmunity [26]. Although we also observed hyper-repopulation in total naïve B cells in our cohort, these cells also contained CD24^hiCD38^hi naïve “regulatory” cells that showed similar kinetics of change, and analyses of changes in CD19+CD20+CD24^hiCD38^hi cells in the absence of CD27 as a marker yielded similar patterns (data not shown). None of the additional B cell or NK measures that we assessed showed significant differences in percentage or kinetics in patients with and without secondary autoimmunity. Perhaps the best evidence of immunological mechanisms underlying secondary autoimmunity in alemtuzumab-treated patients is from two studies published by Jones and colleagues. The first revealed that serum IL-21 was at least 2-fold higher in a cohort of 32 patients who developed autoimmunity compared with 27 patients who did not, a phenomenon that may reflect a genetic predisposition [83]. Although IL-21 was not detected in supernatants from PHA-stimulated PBMC, we did not measure serum IL-21 in this study. The second study provides evidence that homeostatic proliferation of pro-inflammatory CD4+ and CD8+CCR7-RA- or CD45RA+ memory T cells with a restricted T cell repertoire drives the risk for secondary autoimmunity [37]. Although we assessed changes in naïve and memory CD4+ T cell subsets using CD27 instead of CCR7 as a marker, none of them revealed an association with presence of secondary autoimmunity. The identification of a practical immunological biomarker of the risk for secondary autoimmunity remains a high priority.

The repopulation of subsets occurred in relatively distinct temporal patterns, summarized in Table 3. Of special interest are changes that were detected at M5 and sustained throughout the remaining timepoints, or at least to M23, compared with baseline values, as they seem most likely to contribute to the clinical efficacy of alemtuzumab. In addition, changes in several subsets occurred in the majority, if not all patients, including reduced percentages and counts of CD3+ T cells and CD3 + CD4+ T cells, reductions in concentrations of IL-2 and IFN-γ in PHA-stimulated PBMC supernatants, reduced percentages of total memory B cells and increased proportions of naïve “regulatory” B cells.

There are several pitfalls in the current study. The number of M36 and M48 samples was low compared with M0–M23, limiting firm insights into more long-lasting clinically significant changes. Some measures may have been altered in samples shipped overnight, affecting comparisons with studies using fresh samples processed on the same day as collection. The number of samples available for B cell subset analyses was low compared with samples processed after overnight shipment, especially at M17, though the findings are consistent with observations published previously for B cell subsets in alemtuzumab-treated MS patients [22, 26, 31, 56]. Treg function was assessed in an unconventional assay involving rebound proliferation in PBMC depleted of Tregs, making it difficult to compare results with those in which classical proliferation or pro-inflammatory cytokine inhibition assays were employed. In addition, the identity of proliferating cells in CD25-depleted PBMC, and the possibility that depletion of CD25+CD8+ T cells, B cells, and NK cells might affect the assay were not assessed. The current study was designed to survey longitudinal changes in primary T cell, B cell, and NK cell subsets and did not include deep phenotyping and functional assessments in purified CD4+, CD19+, or CD56+ lymphocytes. This is important, since multiple subsets of regulatory CD4+ Tregs [70], CD8+ [80] regulatory B cells [74], and CD56^bright NK cells [89] with more specific and distinct surface phenotypes have been described, as is the case for naturally heterogenous effector lymphocyte subsets [90]. Finally, the study did not include healthy controls or untreated MS patients, which may have provided an additional perspective on the implications of the findings.

The efficacy of currently approved disease modifying treatments for MS is largely linked to the ability to deplete, suppress, or modify pathogenic T and B cells [2]. However, therapies that deplete both T and B cells also result in expansion or unmasking of regulatory cell types, most evident in treatment with cladribine [11], hematopoietic stem cell transplantation [91], and daclizumab [5]. A similar phenomenon occurs in alemtuzumab-treated renal transplant patients [92]. It is tempting to speculate that these findings support the possibility that regulation is a default state in the immune system that can be revealed during recovery from lymphopenia, and in part, recapitulates the developmental regulatory bias known to occur in the prenatal or neonatal periods (reviewed by Simon and colleagues, [93]).

The data in this communication indicate that the immune environment in alemtuzumab-treated multiple sclerosis patients with relapsing-remitting disease is biased in favor of regulatory lymphocyte subsets that are not restricted to the T cell compartment, but also involve robust increases in the percentages of phenotypically regulatory B cell and NK cell subsets. Clinical efficacy may reflect the sum of interactions among these subsets to provide more effective control of multiple pro-inflammatory effector cell types. The data also indicate that CD4+ and CD8+ T cells that express chemokine receptors are enriched in the repopulating lymphocyte pool. Although clear immunological signals of efficacy or incidence of secondary autoimmunity were not identified, exploratory random GEE Poisson analyses support the possibility that the relative balance of several key potentially pathogenic and regulatory subsets is associated with the risk for or protection from inflammatory Gd+ lesions, while the risk for T2 lesions is associated with CD3+CD8+CXCR3+ T cells. Thus, the study identifies biomarkers with potential to predict future disease activity or stability. Future studies are needed to validate these findings and to refine our understanding of the mechanisms of action and risk for secondary autoimmunity in alemtuzumab-treated patients, as well as immunopathogenesis in MS.