A historical perspective

For a long time it was believed that the underlying defect of autoimmune diseases resided in the HSC. Early on, it was noted that haematopoietic radiation chimaeras between rodent strains that were susceptible or resistant to autoimmune disease sometimes developed disease and sometimes not, and that this propensity was generally dictated by the genotype of the bone marrow.4 Thus it made little sense to use autologous HSCT in the treatment of such diseases. This notion was challenged in the early 1990s, when it was demonstrated that rats with adjuvant arthritis responded just as well to autologous or syngeneic bone marrow transplantation as to grafting with allogeneic bone marrow.5 Soon thereafter, several groups reported on the effects of autologous or syngeneic HSCT for experimental autoimmune encephalomyelitis (EAE).6–10 These studies showed that immunoablation by TBI or high-dose cyclophosphamide followed by HSCT could prevent the development of paralysis in SJL mice or Lewis rats. Post-transplant animals were resistant to reinduction of disease, and histological evidence of inflammation within the CNS was absent.

These early animal studies unmasked limitations of autologous HSCT because the results of autologous HSCT in SJL mice are dependent on disease stage. Animals treated early after disease induction had enduring improvement in disability, while there was no neurological improvement in animals treated in the chronic phase of EAE.11 EAE is an autoimmune demyelinating disease induced by vaccination with myelin epitopes, but if demyelination is due to a persistent CNS viral infection as is the case in Theiler’s murine encephalomyelitis virus-induced demyelinating disease (TMEV), HSCT results in further neurological deterioration from viral hyperinfection of the CNS.12 In clinical practice, neuroinflammatory disease responds to HSCT similar to the autoimmune EAE model and not the virally induced TMEV model.

Based on experiences from such animal studies, Burt et al 13 suggested that autologous HSCT should be tried for aggressive inflammatory MS. However, when Fassas et al 14 performed the first autologous HSCT for MS in April 1995, it was limited to secondary progressive MS for safety reasons. Their experiences with 15 patients were summarised in a seminal paper published in 1997, which set the stage for the coming years. The treatment could ‘be used with relative safety’, and some evidence was found that ‘this kind of therapy can suppress disease progression and reduce disability’.14 In the following years many groups reported outcome from treatment of mainly  secondary progressive MS (SPMS) or primary progressive MS (PPMS) patients in small case series.15–23 With increasing experience from more centres, the results were disappointing and many patients continued to deteriorate, especially with longer follow-up.24 25 This was forthrightly summarised by Burt et al, 18 who reported that HSCT in secondary progressive MS with high expanded disability status scores (EDSS >6.0) was a failure and ineffective in preventing disease progression. At this point very few patients with RRMS were treated, but it was noted that patients with RRMS improved in the EDSS score18 and that the effect on the development of new MRI lesions was profound. In one study, the number of gadolinium-enhancing lesions was decreased from 656 in 18 patients pretherapy, to 7 post-HSCT.26 The turning point came in 2009, when two independent groups reported that HSCT could completely abrogate disease activity in a majority of RRMS patients.27 28 These and similar reports led to the 2012 consensus recommendations that HSCT should be considered as a therapeutic option at second line or beyond for patients with RRMS who deteriorate despite standard therapy.29 As a first, HSCT was approved for treatment of RRMS on the national level by the Swedish Board of Health and Welfare in 2016.30

A large majority of published reports on HSCT for neurological conditions have been made on MS, reflecting the higher prevalence. Prompted by the good treatment responses in MS, HSCT was eventually tried for other neuroinflammatory diseases as well. In CIDP and MG, some case reports of successful outcome were published about 10 years ago; the first report of HSCT for CIDP appeared in 200231 and for MG in 2005.32 Recently a register-based study from the European Society for Blood and Marrow Transplantation (EBMT) was published, reporting outcome of patients with NMO treated with HSCT.33

Multiple sclerosis

Several reports with data from the EBMT registry have been published on HSCT for MS, which highlight some of the difficulties encountered in registry-based studies3 34 35 Validity of data can be put into question since registry data are not vetted by site visits for accuracy of disease stage, experience and certification of physician doing disease scoring. In addition, this group of patients is heterogeneous and has been treated with different regimens and different standard of care guidelines. Finally, the registry includes patients from centres with variable experience with the procedure, where some centres have performed only a handful of transplants, which has been identified as a risk factor for TRM.3 Such elements will confound data interpretation significantly.

In addition to the reports of registry data, a large number of uncontrolled studies have been published over the years. In the initial studies, clarifying outcome and toxicity between trials is complicated due to heterogeneity in disease stage (RRMS, SPMS, PPMS), entry criteria (whether by disease progression or number of relapses), conditioning regimens and treatment guidelines. A summary of these trials was recently tabulated.36 By now, it is established that HSCT has a profound effect on inflammation in MS and that it prevents relapses, new MRI lesions and disability in RRMS to a high degree. Whether HSCT also has a beneficial effect in SPMS or PPMS is at present unknown, and as a consequence we will summarise only the trials concerning chiefly RRMS.

We identified four studies containing at least 10 RRMS patients describing the outcome of a total of 188 RRMS patients (table 1).37–40 Further, we will discuss the recently published The Autologous Haematopoietic Stem Cell Transplantation trial in MS (ASTIMS) study, which at present is the only reported randomised controlled trial.41

Summary

Although head-to-head comparisons with other disease modifying drug (DMDs) are essentially lacking, available data suggest that HSCT for MS is superior to currently approved DMDs. The safety profile of the procedure has been improved in later years, and TRM is now on par with generally accepted procedures such as arthroplasty.63 Other serious adverse events include secondary thyroid disease and ITP, occurring to a lesser degree than with comparable advanced forms of immunotherapy, such as alemtuzumab. In light of the above, the procedure could be recommended for RRMS patients with an unfavourable prognosis despite treatment with alternate approved therapy, who seek treatment at experienced centres. Current data suggest that the effect of HSCT in SPMS and PPMS is moderate at best, and further use should be limited to clinical trials.

Neuromyelitis optica

NMO and NMO spectrum disorders (NMOSD) are autoimmune diseases of the CNS primarily affecting the optical nerves and spinal cord. The presence of IgG autoantibodies to water channel aquaporin-4 (AQP4-IgG) is highly specific for these conditions and potentially pathogenic.64 The prognosis is poor due to respiratory failure and severe visual loss due to optic neuritis.65 Treatment includes immunosuppressive and immunomodulatory drugs such as azathioprine and rituximab, but randomised controlled studies are lacking.

The use of autologous HSCT for treating NMO was first described in a 2010 case report.66 The patient was a 23-year-old woman with a severe and rapidly disabling disease course: vision loss, paraparesis, dysaesthesia and radicular pain. Two years after disease onset, she was treated with HSCT using a BEAM conditioning regimen. At follow-up 12 months after the procedure, the patient was stable with no further relapses and normalised motor and sensory functions. Visual acuity had improved only slightly, but optic nerve atrophy was present before treatment.

In 2014 data from a retrospective multicentre study based on data from the European Group for Blood and Marrow Transplantation Autoimmune Diseases Working Party were presented.67 Sixteen patients (13 women, 3 men) with NMO or NMOSD, refractory to other treatment, underwent autologous HSCT between 2001 and 2011. The median age at NMO diagnosis was 29 (range 10–56) and the median time from diagnosis to transplant was 2 years (range <1–17). The median EDSS at baseline was 6.5 (range 2–8.5) and 10/16 patients had AQP4-IgG at diagnosis. Nine patients were treated with an intermediate BEAM-based conditioning regimen; the remaining seven patients had a lower intensity regimen based on Cy plus ATG or thiotepa-Cy alone. The median follow-up was 47 months (range 20–128). Thirteen patients (81%) had a relapse after a median of 7 months (range <1–57) and nine patients (56%) experienced progression of disability. The estimated relapse-free survival after 3 years was 31% (±12%) and after 5 years 10% (±9%); progression-free survival was 48% (±13%) after both 3 and 5 years. In total, three patients were stable and required no other treatment; the other 13 patients required further treatments after HSCT. One patient died from disease progression 14 months after HSCT. The serostatus of AQP4-IgG was unchanged in the eight AQP4-IgG seropositive patients, who were retested after HSCT.

Three of the patients who relapsed after autologous HSCT in the above-mentioned study were further treated with allogeneic HSCT, and the outcome of two of those patients (one man, one woman) was described by Greco et al in 2014.68 Both patients reached clinical remission and improved in disability; EDSS had decreased from 6 to 3.5 and from 8.5 to 7.5 at follow-up after 4 and 3 years, respectively. No new relapses were observed, and in both patients the AQP4-IgG went from positive to negative after treatment.

Aouad et al 69 reported a single case of a 47-year-old woman with NMO and high titres of AQP4 antibodies who underwent autologous HSCT with a disease duration of 11 years. The patient was given two doses of rituximab, as a part of the conditioning regimen (Cy plus ATG). At follow-up 6, 9 and 12 months after treatment, AQP4-IgG could not be detected; she had no relapse, no clinical progression and no new lesions on MRI.

Currently available data suggest that autologous HSCT can reduce inflammatory activity in NMO in the short term, but a vast majority of patients will relapse within 5 years. Conditioning regimens containing rituximab and allogeneic HSCT may improve prognosis, although the few cases described prohibit any firm conclusions.

Stiff person syndrome

Stiff person syndrome (SPS) is a rare autoimmune central nervous system disorder characterised by muscle stiffness and superimposed painful muscle spasms and frequently by the presence of antibodies against glutamic acid decarboxylase (GAD). Many patients respond to immune modulating therapy, but symptomatic treatment with GABAergic drugs such as benzodiazepines and baclofen may also be helpful. Sanders et al70 reported two cases with severe anti-GAD positive SPS, who underwent HSCT with a high intensity conditioning protocol70. The follow-up time was 56 and 32 months respectively. Both patients achieved clinical remission with a return to premorbid function and neither experienced any unexpected serious adverse events. In one of the patients anti-GAD antibodies decreased from 127 U/mL before HSCT to 87 U/mL two months after the procedure.

Available data provide preliminary evidence that HSCT may be an effective therapeutic option in carefully selected cases of severe, treatment refractory SPS.

Chronic inflammatory demyelinating polyneuropathy

CIDP is a disease targeting myelin in the peripheral nervous system. Approximately 80% of patients will respond adequately to first-line treatment,71 but of those 12%–29% eventually become treatment-refractory.72 73 If not treated efficiently secondary axonal loss will accumulate over time, increasing the level of morbidity and risk of mortality.74 Severe disability despite treatment is seen in 13% of patients and mortality rates vary between 4% and 17%, mainly due to respiratory failure or pulmonary embolism.72 75 Treatment-refractory patients with CIDP are usually treated with second-line immunomodulatory drugs such as cyclophosphamide, ciclosporin, rituximab, azathioprine and methotrexate, although evidence for the efficacy of these treatments is lacking.76 77

The use of autologous HSCT in treatment-refractory CIDP was first reported in 2002 by Vermuelen and Van Oers31, who used a BEAM conditioning protocol. The response was maintained until 5 years post-transplant when symptoms recurred.78 This initial report has been followed by mainly case reports of up to three patients,79–82 with one notable exception. In the as of yet largest documented series of patients with CIDP receiving HSCT, Press et al 83 described 11 patients refractory to first-line treatment, and in 10 of them also to one or more second-line treatments. They received HSCT at four different university hospitals in Sweden between 2002 and 2012. The total Inflammatory Neuropathy Cause and Treatment (INCAT) scale and Rankin scores improved significantly within 2–6 months after HSCT and continued to do so at last follow-up. The motor action potential amplitudes improved rapidly within 4 months after the procedure. Eight of the 11 patients maintained drug-free remission at last follow-up after HSCT, and only three patients experienced a relapse. The procedure was generally speaking safe with no TRM, but adverse events included asymptomatic and symptomatic cytomegalovirus and Epstein-Barr virus reactivation, haemorrhagic cystitis and pancreatitis. Of note, one of these patients received HSCT twice due to a post-HSCT relapse. After the second HSCT the patient had a 3-month episode of fever, bronchitis and elevated liver enzymes probably related to Epstein-Barr and cytomegalovirus reactivation.84

Bregante et al 85 described a patient with Sjögren’s syndrome-associated CIDP, who first responded to HSCT, but got a relapse after 6 months, which was treated successfully with allogeneic HSCT. The patient also developed aplastic anaemia at the time of the neurological relapse.

We could identify reports containing in total 20 patients with treatment-refractory CIDP receiving HSCT at 10 different centres (table 2). Four were treated with the intermediate-intensity BEAM conditioning regimen, and the remainder with low-intensity cyclophosphamide-based protocols. Ninety per cent of the patients improved after the procedure, and 35% experienced a relapse. The most common adverse events (apart from the expected adverse events related to acute toxicity) were asymptomatic and symptomatic cytomegalovirus and Epstein-Barr virus reactivation.

These published cases, which include our own experiences, suggest that HSCT may be a treatment that may induce long-term remission even in severe treatment-refractory cases of CIDP with an overall acceptable risk profile. To improve outcome, treatment should preferably be initiated before irreversible axonal damage has occurred. A more precise estimation of the clinical efficacy of HSCT for CIDP will require a randomised controlled trial.

Other peripheral neuropathies

Antimyelin-associated glycoprotein (MAG) neuropathy is an antibody-mediated demyelinating neuropathy clinically characterised by distal, symmetric, predominantly sensory symptoms. Immunomodulating therapies are often tried, but there is insufficient evidence for their effect.86 In the report by Mahdi-Rogers et al mentioned above,80 the authors also described a patient with demyelinating polyneuropathy with IgM paraproteinaemia and MAG antibodies who first improved after HSCT, but relapsed after 30 months.

Multifocal motor neuropathy (MMN) is an uncommon immune-mediated, slowly progressive, purely motor neuropathy that is characterised by the presence of multifocal conduction blocks in motor nerves and frequently serum anti-GM1 antibodies. IVIg is to date the only therapy with proven efficacy providing transient improvement in muscle strength. A single report describes a patient with MMN who responded to IVIg but had no benefit from treatment with HSCT using a low-intensity conditioning regimen.87

The POEMS syndrome (polyneuropathy, organomegaly, endocrinopathy, monoclonal protein and skin changes) is a paraneoplastic disorder secondary to a plasma cell dyscrasia. The efficacy of HSCT in POEMS syndrome reported in case series has been mainly been based on haematological criteria, and clinical recovery of peripheral neuropathy has not always been specifically evaluated.88 It is beyond the scope of this article to review the extensive literature on HSCT in POEMS syndrome describing more than 100 cases, and the interested reader is instead referred to the recent reviews in the field.89 90

Myasthenia gravis

MG is neurological disease with immune reactivity against the postsynaptic endplate of the neuromuscular junction.91 Apart from symptomatic treatment with cholinesterase inhibitors, many patients also require immunomodulatory treatment. The prevalence of treatment-refractory MG has been estimated to be 10% of patients with generalised disease.92 93

The use of autologous HSCT in MG was first reported in 2004 by Burt et al, 32 who employed a low-intensity conditioning regimen. The patient, a 30-year-old woman with severe therapy-resistant anti-muscle specific kinase (MUSK) antibody positive MG (Osserman grade IIB/III), was free from symptoms (Osserman grade 0) 2 months after HSCT, and prednisolone could be tapered after 4 months. There were no HSCT-related adverse events during the 6-month follow-up.

Bryant et al 94 reported seven cases of severe MG treated with HSCT. Six patients underwent HSCT for MG and one for follicular lymphoma with coincident active MG. Mean ages at MG diagnosis and at HSCT were 37 (±11) and 44 (±10) years, respectively. Five patients had acetylcholine receptor antibodies and two were seronegative. MG severity was graded as moderate (grade III) to life-threatening (grade V) by the Myasthenia Gravis Foundation of America (MGFA) clinical classification, and patients were refractory to treatments including pyridostigmine, corticosteroids, additional immunomodulators and plasma exchange or IVIg. Four patients had previously undergone thymectomy; none had thymoma. All patients received high-intensity conditioning regimens containing TBI or busulfan, leading to durable remission with no residual MG symptoms and freedom from ongoing MG therapy in all patients. The median follow-up was 40 months (range 29–149 months). Acetylcholine receptor antibodies were not analysed after HSCT. Three patients experienced transient viral reactivations and one developed a secondary autoimmune disease after HSCT, all of which resolved or stabilised with treatment.

Håkansson et al 95 described a patient with severe MG, diagnosed at the age of 26, refractory to corticosteroids, four oral immunosuppressants, cyclophosphamide, rituximab and bortezomib. At the age of 63, she was treated with HSCT with the intermediate BEAM conditioning regimen. At follow-up 2 years after the procedure, the patient had significantly improved in objective tests (MG composite score decreasing from 64 to 6). Diplopia was her only remaining symptom and she was off all medications for MG. Interestingly, acetylcholine receptor antibodies were still detectable after HSCT. No serious adverse events were reported.

A single report of one patient with intractable MG undergoing allogeneic HSCT describes a treatment-free remission with minor residual symptoms at 40 months after transplantation.96 The role of allogeneic HSCT in a disease as MG should be limited due to its high risk of mortality and the possible occurrence of secondary MG as a manifestation of graft versus host disease.

Available data provide preliminary evidence that HSCT can be an effective therapeutic option in carefully selected cases of severe, treatment-refractory MG.

Concluding remarks

In this review, we have summarised the past 20 years of experience with autologous HSCT for neurological diseases. The vast majority have been patients with MS, and by now more than 1000 individuals have been treated with this procedure. Available data are very consistent on ‘no evidence of disease activity’, and 68%–70% of RRMS patients will remain in NEDA 5 years after HSCT regardless of conditioning regimen. Safety has been improved and the TRM has been estimated at 0.3% in the last decade. These numbers compare favourably with conventional treatment and open up for a more widespread use. Although considerably fewer patients with MG and inflammatory neuropathies have been treated so far, the reported outcome looks promising and should be followed by formalised clinical trials. Meanwhile, the procedure could be considered in patients with severe treatment-resistant disease. In contrast, HSCT does not seem to be able to contain disease activity in NMO, and further use outside of clinical trials is questionable at best.

Review criteria

A search for original articles was performed in Medline and PubMed. The search terms ‘multiple sclerosis’, ‘neuromyelitis optica’, ‘chronic inflammatory demyelinating polyneuropathy’, ‘multifocal motor neuropathy’, ‘inflammatory neuropathy’, ‘peripheral neuropathy’, ‘myasthenia gravis’, ‘hematopoietic stem cell transplantation’, ‘relapsing-remitting’ and ‘clinical trial’ were used alone or in combination. The reference lists of identified articles were scrutinised for further relevant papers.

For efficacy in MS, we included only those reports that included at least 50% RRMS patients and reported the outcome of at least 10 patients. Four studies were found using this strategy. In addition, we included the only randomised controlled trial that was identified, although it mainly contained data on SPMS patients.

For NMO we included all relevant reports found using this strategy.

For SPS we included all relevant reports found using this strategy.

For CIDP and other inflammatory polyneuropathies we included all relevant reports found using this strategy.

For MG we included all relevant reports found using this strategy.