Safety of Alemtuzumab and Autologous Hematopoietic Stem Cell Transplantation Compared to Noninduction Therapies for Multiple Sclerosis

We performed a nationwide register-based cohort study of alemtuzumab and AHSCT compared to matched noninduction therapies, linking data from the Swedish MS register to national health care and demographic registers.

We identified 132 alemtuzumab-treated and 139 AHSCT-treated (68% Cy/ATG, 32% BEAM/ATG) patients with mean follow-up of 3.5 and 4.2 years, respectively, in analyses of death, new therapy, and prescription drug–based outcomes and 2.8 and 3.4 years, respectively, for the diagnosis-based outcomes and cancer. The 2,486 noninduction therapies (36% natalizumab, 29% dimethyl fumarate, 22% rituximab, 13% fingolimod) were matched to the alemtuzumab and AHSCT therapies in a reference group with a mean follow-up of 4.3 and 3.6 years, depending on the outcome.

In this observational study using high-quality data from clinical practice recorded in Sweden's nationwide health registers, we investigated safety outcomes for all patients with MS in Sweden treated with alemtuzumab and AHSCT compared to a matched reference group receiving noninduction MS DMTs. While the effectiveness of both alemtuzumab and AHSCT is considered high, safety concerns have limited their use, and studies comparing safety outcomes across therapeutic options have so far been rare. Our main findings were a higher incidence of thyroid disease with alemtuzumab and, to a lesser extent, with AHSCT, a higher incidence of infection with AHSCT, and a low incidence of nonthyroid autoimmune disease for both therapies.

A major concern with alemtuzumab treatment has been an increased risk of secondary autoimmune disease, in particular thyroid disease and immune-mediated thrombocytopenia.^2,11,12 We report here a higher incidence of thyroid disease after treatment with alemtuzumab than previously seen in the clinical trials^2,11,12 and in line with recent observational data.¹³ Notably, preexisting autoimmune conditions represented an exclusion criterion in the Comparison of Alemtuzumab and Rebif® Efficacy in Multiple Sclerosis (CARE-MS) trials.^2,11 We further found a higher incidence of thrombocytopenia after alemtuzumab compared to the matched reference group, similar to what was found in the clinical trials.^2,11,12 The incidence of first-ever nonthyroid autoimmunity with alemtuzumab was similar to that in the reference and AHSCT groups. Unfortunately, the groups were too small to analyze whether preexisting or posttreatment emerging thyroid autoimmunity was associated with a higher incidence of nonthyroid autoimmune disease after treatment. No cases of autoimmune hepatitis or hemophagocytic lymphohistiocytosis were recorded. The incidence of diagnosed infection in the alemtuzumab group was similar to that in the reference group, but the use of antibiotics and especially antivirals was higher after alemtuzumab, likely reflecting an actual or perceived increased infection risk that is managed prophylactically. One of the 4 deaths in the alemtuzumab group was directly related to the alemtuzumab treatment, a case of cytomegalovirus reactivation with multiple organ failure occurring within 1 month of the first treatment cycle. For the remaining 3 cases, the relation to alemtuzumab was uncertain.

A range of different treatment-related adverse events have been reported for AHSCT, of which treatment-related mortality is the most severe.³ In our study, only 1 death occurred after AHSCT during the observation period, a suicide 6 years after therapy start. The mortality rate for AHSCT was similar to that of the matched reference group and much lower than in most previous reports.³ This may be explained in part by the type of ablative conditioning used because less aggressive regimens have become more common. In this cohort, Cy/ATG had been used in 68% and BEAM/ATG in 32% of the patients undergoing AHSCT, and there were some differences in the baseline characteristics of these subgroups (table 2). A diagnosis of infection was the most common adverse event with AHSCT. The incidence was highest immediately after therapy start but remained higher than for both the alemtuzumab and the matched reference groups after 6 months from therapy start. The incidence of infection was higher close to BEAM/ATG conditioning compared to Cy/ATG but dropped to similar levels after 6 months. This difference between the conditioning regimens should be interpreted cautiously because routines for infection prophylaxis may have differed. According to previous studies, the infection risk in the reference group varies depending on type of DMT, with rituximab (22% of our reference group) having the highest risk of hospitalized infections.¹⁴ The incidence of thyroid disease in the AHSCT group was higher than in the matched reference group but lower than in the alemtuzumab group. Disorders relating to the reproductive organs and fertility were common in the AHSCT group, especially in women and in those having received Cy/ATG conditioning. The Cy/ATG subgroup was also more likely to start a new therapy compared to the BEAM/ATG subgroup.

The main strengths of this study are the use of national health registers with mandatory reporting and virtually complete coverage, together with the Swedish MS register with its high-quality data on treatment episodes and the availability of a large reference group of noninduction therapies. Although the data were of good overall quality, there was significant missingness in the rating scales of disease severity around treatment start, which was distributed unevenly across therapies. However, there were no missing observations for any of the outcome variables, and the missingness affected mainly the ability to perform more exact matching on disease characteristics. We also had no information on visits and diagnoses in general-practice outpatient care, and it is therefore likely that the incidences reported for the diagnosis-based outcomes are underestimated. This is a concern mainly for milder adverse events managed in primary care because more serious illness will be treated in inpatient or specialized outpatient care for which data were available. Because the follow-up ranged between 3 and 4 years, it was not possible to detect longer-term adverse events, which is especially important to consider when interpreting the incidence of cancer and serious late-onset autoimmune events. Additional limitations include potential channeling to therapy and surveillance bias, making direct comparisons between the groups more difficult. However, the regional preference and experience of treatment with either alemtuzumab or AHSCT seemed to be the decisive factor explaining the difference in treatment choice (table 2). The channeling to therapy effect would therefore be strongest in comparisons to the noninduction therapy reference group, for which matching was performed to mitigate this issue. It is likely that residual bias still exists in this comparison because the matching did not involve any MS-specific covariates and it is well accepted in the clinic that alemtuzumab and AHSCT are generally used to treat more aggressive disease, normally requiring breakthrough on previous therapies (as can be seen in the number of previous therapies in table 2). The difference in aggressiveness of MS would be more important if we were to investigate effectiveness outcomes, but the higher use of previous MS therapies could potentially also affect safety outcomes by a carryover effect, in which a previous therapy is actually causing the events that we find. However, because the majority of previous therapies are those in the reference group, any effect that could carry over to the alemtuzumab and AHSCT groups would also be seen in the reference group. A possible source of surveillance bias was that the patients undergoing AHSCT will generally be monitored more closely during and in the months after conditioning, often in the hospital, compared to those receiving alemtuzumab or the noninduction therapies. On the other hand, alemtuzumab-treated patients are followed up more carefully with blood and urine tests. These differences also complicate the severity assessment of infections, especially around therapy start for AHSCT because hospitalization occurs regardless. In addition, prophylactic antibiotics and antivirals are recommended for 6 to 12 months after AHSCT and antivirals for 1 month after alemtuzumab infusion, limiting the use of these as indicators of infectious events and potentially altering the comparison to the reference group for which there are no such recommendations. Finally, the preference of conditioning in the AHSCT group has changed over time, with a sharp increase of Cy/ATG in 2014 (data not shown). Because recommendations and guidelines for AHSCT treatment have likely also changed over the years, comparisons between the conditioning subgroups should be made with caution. Studies focused specifically on comparing these conditioning regimens are needed to explore any differences further.

Despite the descriptive nature of this study, the differences in incidence of thyroid disease and infection between the groups were great enough to allow a comparative interpretation even in the absence of confounder adjustment. There were differences in baseline characteristics between the groups, but they were mostly minor, and it is unlikely that confounding alone could cause the sharp increase in incidence that occurred directly after patients received treatment. The size of the differences and the clear temporal relationship with therapy start make a causal interpretation plausible.

A third induction-type therapy, cladribine, was approved for MS in 2017, but because it is so new, there were not enough treated patients to include it in our analyses. Future studies are needed to investigate how cladribine compares to alemtuzumab and AHSCT.

Collectively, in this study of 132 alemtuzumab- and 139 AHSCT-treated patients, we confirm the previously reported increased incidence of thyroid disease for alemtuzumab and, to a lesser extent, also for AHSCT and found a higher incidence of infection for AHSCT (highest for BEAM/ATG) compared to both the alemtuzumab group and the matched noninduction therapy reference group. The incidence of first-ever nonthyroid autoimmune events was similar between the groups. Disorders relating to reproductive organs and fertility were common in the AHSCT group, especially in women. Other adverse events were rare but still more common in the alemtuzumab and AHSCT groups than in the matched reference group. Overall mortality was slightly higher in the alemtuzumab group, but only 1 of the 4 deaths was clearly linked to the treatment.