Alemtuzumab (ALZ) is an immune reconstitution therapy for treating relapsing-remitting multiple sclerosis (RRMS). After two ALZ courses of 5 and 3 days, administered 12 months apart, new ALZ infusions can be administered at signs of new inflammatory disease activity. Each ALZ infusion causes a pan-lymphocyte depletion, where all circulating lymphocytes are reduced within a few days. After this pronounced lymphocyte reduction, T and B lymphocytes repopulate, which can contribute to a sustainable effect (1, 2). In clinical practice, ALZ is mostly used as a second- or third-line treatment for RRMS, and this treatment is less restricted, in terms of patient age and comorbidities, than the ALZ treatments described in the clinical trial program of ALZ (3–5). Furthermore, these studies (3–5) excluded patients with clinically significant autoimmunity or thyrotropin receptor antibodies (TRAbs) at the baseline. Hence, “real-world” patients treated with ALZ differ significantly from the RRMS populations that were included in the ALZ treatment trials.

Alemtuzumab is considered a highly effective disease-modifying therapy for RRMS. However, its utility is limited because it increases the risk of infections, infusion-related side effects, and in particular, secondary autoimmune diseases (SADs). Among patients treated with ALZ, 30–48% develop SADs, and the incidence peaks after 2–3 years (3–8). The most common SADs are autoimmune thyroid disease (AITD), immune thrombocytopenia (ITP), and autoimmune nephropathies. In the pivotal ALZ studies CARE-MS I and II (4, 5) and their extensions (7, 8), the prevalence of AITD was 20 and 40%, ITP was 1 and 3%, and autoimmune nephropathies were 0.3 and 0.3%, respectively. A range of other rare disorders has also been reported in patients treated with ALZ, including neutropenia, hemolytic anemia, and autoimmune hepatitis (4, 5, 7–9). A recent safety analysis of the ALZ clinical trial program found that post-ALZ treatment SADs were similar among patients with preexisting non-multiple sclerosis (MS) autoimmunity (35.4%) and those without it (35.3%), and these SADs were serious in 8.8 and 9.1% of cases, respectively (10). In 2019, ALZ underwent a procedure under Article 20 of Regulation (EC) No 726/2004 resulting from pharmacovigilance data, which led to label change effective from January 2020 (11). However, since the beginning of the last year, the restrictions on the use of ALZ as first-line therapy became less strong. According to the last summary of the product characteristics update from the European Medicines Agency, ALZ can now be used to treat RRMS if the disease is highly active despite treatment with at least one disease-modifying treatment or in patients with rapidly evolving severe RRMS (12). ALZ must also no longer be used in patients with certain heart, circulation, or bleeding disorders or in patients who have autoimmune disorders other than MS.

The pathogenic mechanism for post-ALZ SADs remains unclear, but SADs appear to occur during and after a phase of immune reconstitution, following lymphopenia. Similar inductions of autoimmunity have been reported after antiretroviral therapy in patients with human immunodeficiency virus, and in patients after bone marrow transplantation (13, 14). Although the association between lymphopenia and autoimmunity is well recognized, most patients with lymphopenia do not develop autoimmunity during lymphocyte recovery; hence, additional factors seem to be involved in this process.

It was previously suggested that after ALZ treatment, SADs occurred in patients with increased pre-treatment serum interleukin-21 levels, which gave rise to elevated T-cell apoptosis and cell cycling (15). However, those findings could not be verified in larger prospective cohorts with currently available interleukin-21 immunosorbent assays (16). A small retrospective study on the use of custom-made TRAb assays demonstrated that measuring the levels of TRAb at the baseline and antibodies (Abs) against thyroid peroxidase (TPOAbs) could predict an increased risk of AITD after ALZ treatment (17). Furthermore, a prospective study employed standard tests for TPOAbs and thyroglobulin Abs (TgAbs); they demonstrated that the detection of these autoimmune Abs (auto-Abs) at the baseline indicated an increased risk of developing clinically manifest AITD after ALZ treatment (18). Recently published clinical trial data support a risk of SAD among 51–67% and 26–35% of patients with respective without TPOAbs at the baseline (10). However, lymphocyte repopulation dynamics have not been shown to predict SADs occurrence after ALZ treatment (19). A recent study presented findings supporting that the risk of SADs in predisposed ALZ-treated patients was linked to ALZ-specific immune repertoire changes, including the restriction of T- and B-cell repertoire, reduced thymopoiesis, homeostatic proliferation, and disparate dynamics of clonal T- and B-cell expansion (20). Furthermore, previous treatment and the treatment sequence impact the efficacy and safety profile of ALZ (21).

The present observational, prospective, nationwide study included a “real-world” Swedish cohort of patients with RRMS that were treated with ALZ. We explored whether we, in line with previous studies in the field, could confirm that the detection of auto-Abs at the baseline (i.e., before initiating ALZ) or during ALZ treatment could predict the development of SADs. This prediction could improve monitoring safety and might have implications for therapeutic decisions.

We included a total of 124 (74 female subjects) patients with RRMS, of which three patients participated in CARE-MS I (4) and its extension study TOPAZ (22). A total of 77 patients were recruited from the MS Center of Sahlgrenska University Hospital, Gothenburg, and 47 patients from the “Immunomodulation and Multiple Sclerosis Epidemiology” study, at the Karolinska Institute, Stockholm, Sweden.

Most patients (n = 85, 69%) switched to ALZ from other highly effective disease-modifying therapies (natalizumab, rituximab, and fingolimod), 18 (14%) patients switched from a moderately effective therapy (dimethyl fumarate, teriflunomide, interferon beta, and glatiramer acetate), and 21 (17%) patients were treatment-naive. Patients were switched to ALZ when a disease breakthrough occurred during an ongoing disease-modifying therapy (n = 57), when JC-virus was detected in patients treated with natalizumab (n = 30), when adverse events occurred (n = 12), or for other reasons (e.g., family planning and patient request; n = 4). The median follow-up time was 4.5 years (ranging from 1.3 to 11.2). Baseline patient demographic and clinical characteristics are presented in Table 1.

In this nationwide consecutive series of patients with RRMS that initiated ALZ, we observed that SADs and AITD developed in 24 and 21% of patients, respectively, at 24 months, and in 44 and 40% of patients, respectively, during a median follow-up of 4.5 years, consistent with previous reports (4, 5, 7, 8). Thyroid auto-Abs were detected in 62% of patients with AITD, and the presence of TRAbs at the baseline increased the risk of AITD by 50%. HR for the presence of each thyroid auto-Ab and for subsequent clinical AITD increased during follow-up and was 4.6 at 24 months for TRAb. Patients that developed thyroid auto-Abs had a significantly higher risk of AITD, and they experienced a significantly earlier AITD onset compared to patients without thyroid auto-Abs. In contrast, non-thyroid SADs were not associated with the development of non-thyroid auto-Abs.

In our study population, the different specific thyroid auto-Abs were relatively evenly distributed at baseline. This finding contrasted with findings from previous studies (17, 18, 33). Compared to our findings, two of these studies showed a higher prevalence of TPOAbs and TgAbs (18, 33), and a third study showed a slightly higher proportion of TRAb positivity, before initiating ALZ treatment (17). Data from the National Health and Nutrition Examination Survey III (34) also show a higher prevalence of TPOAbs (11.3%) and TgAbs (10.4%) in the general disease-free population than those observed at baseline in our study population (3% each), which may be due to the younger age of the MS population. In the general population, TPOAbs were associated with thyroid disease, but TgAbs alone (in the absence of TPOAbs) were not (34). In contrast, the prevalence of TRAb is close to zero in healthy persons (35) and their presence seems to be crucial in the pathogenesis of AITD (36).

The prevalence of ANA in our study population was 12% at baseline and it remained essentially unchanged during follow-up. This elevated prevalence of ANA was consistent with that found in a previous report (37), and it was higher than the prevalence found in the general population (38). Consistent with previous surveys, we did not find an elevated risk of systemic lupus erythematosus during follow-up (37). Some previous studies found that ANAs occurred at a higher frequency among patients with MS than among controls, probably reflecting an autoimmune predisposition supported by the increased risk of other autoimmune diseases reported in MS populations (38).

Our findings suggested that screening for thyroid-Abs before initiating ALZ might improve the selection of patients appropriate for ALZ treatment. At the baseline, nine of our patients had thyroid auto-Abs, but only those with TRAbs had a significantly higher risk of developing AITD. This association was consistent with the results of a previous study (17), which showed that TRAbs (measured with custom-made assays) could be detected prior to any changes in thyroid function in up to one-third of patients with post-ALZ AITD. Moreover, the presence of TRAbs prior to ALZ treatment was strongly predictive of subsequent AITD. In contrast, patients with TPOAb and TgAb positivity at the baseline did not show an increased risk of clinically manifesting AITD at follow-up. Similar results were reported in a previous study, which concluded that TPOAb alone had a limited value in risk stratification for post-ALZ AITD (33).

Thyrotropin receptor antibodies are considered pathognomonic and pathologic, and they were detected in the vast majority of patients with Graves' disease (36, 39). However, the functional consequence of TgAbs remains unclear because they do not cause thyroid cell destruction (39). In contrast, TPOAb may be cytotoxic to thyrocytes in autoimmune thyroiditis (i.e., Hashimoto's thyroiditis), but their role has not been established in Graves' disease (40). In patients treated with ALZ, Graves' disease was the most common AITD (41). This finding might explain why, among patients treated with ALZ, the role of TRAb in the risk of thyroid disease was more coherent and predictive than the roles of other thyroid auto-Abs.

The majority of patients with AITD in our study population were diagnosed with Graves' disease and they were treated according to standard procedures as previously reported in a meta-analysis (41). However, opinions diverge on the severity of ALZ-induced thyroid disease. Some authors have suggested that the disease course was more favorable in ALZ-induced Graves' disease than in conventional Graves' disease. Indeed, the majority of patients with ALZ-induced Graves' disease were successfully managed with medical treatment alone (anti-thyroid drug and/ or levothyroxine) or the disease spontaneously remitted (42). In contrast, other authors have suggested that ALZ-induced Graves' disease has unique, complex features, with an unpredictable, fluctuating course (23, 33), which frequently requires definitive treatment (i.e., radioactive iodine or thyroidectomy) (33, 41).

The incidence of ALZ-induced SADs other than AITD was considerably lower than the incidence of AITD. ITP was the second most frequent SAD; it represented 1% of SAD cases. These findings were consistent with those from the pivotal (4, 5) and extension studies (7, 8), on ALZ. Moreover, we did not find any of the autoimmune disorders described in post-marketing reports of ALZ, e.g., autoimmune hepatitis, primary biliary cholangitis, or hemophagocytic lymphohistiocytosis (9, 43). Importantly, non-thyroid SADs were not associated with the development of non-thyroid auto-Abs. Finally, we did not detect any signs of auto-Abs associated with the rare autoimmune conditions that prompted the new indication and safety monitoring for ALZ (44, 45).

It was suggested that changes following ALZ, during the immune reconstitution period (i.e., the rapid repopulation of immature B lymphocytes and homeostatic T lymphocyte proliferation), might be involved in the development of AITD (1, 14, 46, 47). It remains unclear whether these changes are connected to the beneficial effects of ALZ in RRMS. We found that patients that had developed a SAD or AITD had slightly lower EDSS scores at follow-up compared to those that did not develop a SAD. However, this finding was not supported by reductions in the rates of relapse or lesion formation on MRIs. In contrast, a previous study showed that both disability and disease activity were reduced in patients with AITD post-ALZ compared to patients without AITD (48). Estimating disability with EDSS at specific time points is less reliable compared to CDW due to intra- and inter-rater variability and to the short-term impact of relapses (12, 49, 50). Thus, the differences we found in EDSS should be interpreted with caution because they were not supported by CDW or NEDA-3.

The risk of developing a SAD was higher among current and previous smokers than among patients that had never smoked, but the risk was not affected by the presence of an autoimmune disease other than MS at baseline. Previous studies have shown that both tobacco smoking and autoimmunity were associated with several autoimmune diseases (51–54), including MS (55). However, reports on the risk of ALZ-induced autoimmunity in smokers have shown conflicting results. One previous study reported no significant association between baseline smoking status and AITD (18); conversely, another study demonstrated an odds ratio of 3.05 for AITD in patients with a smoking history at the baseline (56). Smoking was previously shown to have a broad immunologic impact. Smoking altered T- (57) and B-lymphocyte functions (58), upregulated complement factors (59), and modified gene expression (60). All or some of these effects may be important in the development of SADs post-ALZ. Thus, it would be reasonable to consider smoking a potential risk factor for SAD development post-ALZ and to incorporate this potential association in the counseling process prior to treatment.

This study had some limitations. First, although this study was a nationwide survey, only a limited number of patients were treated with ALZ in Sweden. Second, most patients received two cycles of ALZ, but three patients received only one cycle and were followed for a shorter time (the shortest follow-up time was 1.3 years). Thus, because post-ALZ SADs mostly develop within 2–3 years from the baseline, we may have missed a few isolated cases of SAD. Third, we conducted a retrospective review of medical records to obtain information on the association between AITD or non-thyroid SADs and smoking habits or concurrent autoimmune disorders other than MS in the study population. The retrospective nature of this review might have introduced the risk of unmeasured bias, and other data sources than medical records might have improved data accuracy. Furthermore, the prospective observational study design makes dropouts inevitable. Fourth, there was a 3-month interval between auto-Ab tests compared to monthly safety tests, including those for SADs. This difference would increase the likelihood of an early diagnosis of SADs compared to early detection of a prior positive auto-Ab test. However, the impact of this difference seemed limited since almost all patients in the frequently monitored subgroup developed thyroid disease after the detection of auto-Abs. Finally, only a few patients developed non-thyroid SADs. However, the results were consistent and probably reflect immune mechanisms for ALZ-induced non-thyroid SADs that are different from the mechanisms associated with thyroid autoimmune diseases.

In conclusion, this national study confirmed that patients treated with ALZ for RRMS had a high risk for SADs dominated by AITDs. We showed that thyroid auto-Abs often preceded the development of thyroid disease. This finding, concerning essentially TRAbs, can probably be used to identify individuals at high risk of AITD, before initiating and during ALZ treatment. We also confirmed that tobacco smoking increased the risk of SADs. In contrast, the risk for non-thyroid SADs was low, and monitoring non-thyroid auto-Abs did not seem to provide any additional information for predicting non-thyroid SADs.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The study involved human participants and was reviewed and approved by the Regional Ethics Review Board in Gothenburg, Sweden (Reference number 460–13). The Immunomodulation and Multiple Sclerosis Epidemiology study was approved by the Regional Ethics Review Board in Stockholm, Sweden (Reference number 2011/641-31/4). The patients/ participants provided their written informed consent to participate in this study. The study conformed with the Code of Ethics of the World Medical Association (Declaration of Helsinki) (61).

SS: data analysis and interpretation, data acquisition, and manuscript drafting. LN and MA: data acquisition and critical revision of the manuscript for intellectual content. FA: data acquisition, method development, and critical revision of the manuscript for intellectual content. IK and TO: study concept and design, data acquisition, and critical revision of the manuscript for intellectual content. CM: study concept and design, data acquisition, method development, and critical revision of the manuscript for the intellectual content. JL: study concept and design, critical revision of the manuscript for the intellectual content, manuscript drafting, study supervision, and funding acquisition. All authors contributed to the article and approved the submitted version.