Autologous HSCT is the most widely used form of HSCT. In hemato-oncological conditions, it is a relatively safe procedure with a TRM typically below 3%. Toxicities and transplant-related causes of death include sepsis, cytomegalovirus infection, and haemorrhage. The overall TRM for autologous HSCT in AD now is approximately 7%, although it was as high as 23% in one of the first pilot studies 21. In AD, diagnosis and extent of organ involvement, age, and comorbidity are patient-related determinants of toxicity and TRM (Figure 3). TRM and toxicity also depend on the conditioning regimen and whether or not TBI is performed 22. With adaptation of eligibility criteria (for example, exclusion of patients with severe pulmonary hypertension) and modifications of transplant regimens (for example, lung shielding with TBI), complications from HSCT can usually be managed in experienced hands, and TRM has dropped as a consequence. It was less than 1% for non-TBI nonmyeloablative, less than 2% for low-intensity myeloablative, and 13% for high-intensity myeloablative regimens 23. Compared with TRM, efficacy seems less influenced by intensity and type of conditioning, although this may be confounded by the severity of underlying disease. In SLE patients, nonmyelo-ablative conditioning had an efficacy comparable to that of myeloablative conditioning with a markedly lower TRM of 2% versus 13% 124. A similar observation was made for SSc where nonmyeloablative regimens had a TRM of less than 4% in contrast to 23% for myeloablative conditioning with TBI, with similar efficacy 212526. Such comparisons, however, are not based on prospective controlled trials and must therefore be interpreted with caution.

Nonmyeloablative autologous HSCT in SLE patients resulted in a disease-free 5-year survival of 50% and an overall 5-year survival of 84% with improvements of SLEDAI (Systemic Lupus Erythematosus Disease Activity Index) scores and ANA, anti-double-stranded DNA, and complement levels, as shown in a single-centre study 24. A retrospective multicentre analysis of 53 patients of the EULAR/EBMT registry showed an improvement of SLEDAI scores in 66%, but one third relapsed subsequently (after 3 to 40 months) 1. Autologous HSCT also proved effective in antiphospholipid syndrome: 10 of 22 patients discontinued anticoagulation therapy, 78% of whom remained without further thrombotic events 27. Interestingly, soft tissue calcifications resolved after autologous HSCT 28.

In patients with SSc, the 5-year event-free survival after HSCT in the North American and European studies was 64% 2126. Autologous HSCT resulted in a remarkable reversal of skin thickening, quality of life, and stabilisation of organ function 2526. A significant decrease in the modified Rodnan skin score was achieved in 73% of 26 patients after 1 year and in 94% after 5 years. Relapses of SSc occurred in a third of the cases after 2 to 4 years 326. Stabilisation of heart, lung, and kidney function was seen after autologous HSCT. Persistent changes in capillary structure have been described recently after autologous HSCT in seven patients 2930. A comprehensive study on vascular markers in skin biopsies before and after transplant suggested that HSCT induces neoangiogenesis 31.

Three prospective studies are ongoing to investigate the safety and efficacy of nonmyeloablative conditioning followed by autologous HSCT compared with monthly intravenous pulse cyclophosphamide in SSc patients: the Autologous Stem Cell Transplantation International Scleroderma (ASTIS) trial in Europe and the American Scleroderma Stem Cell versus Immune Suppression Trial (ASSIST) and the Scleroderma: Cyclophosphamide or Transplantation Trial (SCOT) in North America. In the SCOT trial, a TBI-based myeloablative approach is used whereas ASTIS and ASSIST use a lympho-depleting regimen with ATG. At the time of this writing, 124 patients have been randomly assigned in the first of these trials (ASTIS trial). Long-term follow-up will be needed in all trials to evaluate durability of responses and late toxicities.

In RA, autologous HSCT was analysed in several studies. A registry analysis of 76 cases showed an American College of Rheumatology 50% improvement in 67% of patients and a significant reduction in the level of disability 3233. However, the relapse rate was high and disease-modifying antirheumatic drugs (DMARDs) had to be reintroduced in most of the patients. Interestingly, sensitivity to DMARDs seemed restored after HSCT. The availability of effective biologicals to treat severe RA has led to a reduced demand for intensive therapies such as (autologous) HSCT. Similar to RA, autologous HSCT showed a response rate of 68% in severe JIA. However, 5 of 20 responding patients relapsed, and a significant TRM including some cases of macrophage activation syndrome was reported 34. Two case series showed positive effects of autologous HSCT in systemic vasculitis. Four patients were treated with nonmyeloablative stem cell transplantation (two with neurovascular Behçet disease, one with neurovascular Sjögren syndrome, and one with Wegener granulomatosis), and three remained in complete remission after 2 years 35. One patient with Behçet disease did not improve. A further study reported an 86% response in 14 patients with different types of vasculitis. The median duration of response was 45 months. Two of the patients relapsed; no TRM in these patients was reported 36.

High-dose cyclophosphamide conditioning without stem cell rescue has been tested in SLE and scleroderma patients 373839. The rationale for that procedure is a rapid eradication of autoreactive effector cells while sparing stem cells because their high content of aldehyde dehydrogenase inactivates cyclophosphamide metabolic products. In SLE, 5 of 14 patients had a complete response, 6 had a partial response, and no TRM was reported. In 6 SSc patients, 1 patient died of TRM and 2 relapsed, whereas 3 had a sustained response after 12 to 24 months. These observations suggest that the key therapeutic effect from autologous HSCT is achieved by high-dose cyclophosphamide but do not exclude a role for HSCT as the long-term effects of this regimen are unknown. In addition, due to safety concerns, this variant protocol has not been adopted in other centres.

In allogeneic HSCT, HSCs are collected from related (HLA-matched) siblings or matched unrelated donors. Different conditioning regimens are employed, usually involving two cytotoxic agents (for example, fludarabin + busulphan), and ATG with or without TBI. Post-transplant immunosuppression is given to prevent GVHD, typically with cyclosporin or methotrexate. Allogeneic HSCT is the only therapy that offers the prospect of a cure for AD, due to its ability to induce a graft-versus-autoimmunity effect, but its attendant risk of GVHD precludes its routine use in AD. Long-term remissions have been observed in several patients with AD as well as in patients with concomitant hematologic malignancy who relapsed or did not respond to autologous HSCT 40. In 23 patients with aplastic anemia and concomitant AD, the survival after allogeneic HSCT in complete remission of AD was 64% at 13 years; TRM was 21.7%. In two patients with concomitant organ-specific autoimmunity (one with autoimmune thyroiditis and one with type 1 diabetes), AD did not resolve but favourable responses were seen in 21 patients suffering from RA, SLE, discoid lupus, eosinophilic fasciitis, and psoriasis. The studies suggest that allogeneic HSCT leads to a high relapse-free survival in patients with AD and concomitant hematologic disease. Two SSc patients treated with allogeneic HSCT showed improvement of skin thickening and resolution of abnormalities on high-resolution computed tomography and/or bronchoalveolar lavage 41. Both received myeloablative conditioning and developed full donor chimerism. One patient developed chronic GVHD necessitating immunosuppression and died after 18 months due to pseudomonas sepsis. In two other SSc patients, a chimerism of 10% to 15% after nonmyeloablative conditioning was associated with a sustained complete remission for over 3 years without GVHD 4243. In the only reported case of nonmyeloablative conditioning plus allogeneic HSCT in RA, mixed chimerism resulted in marked amelioration of RA, without GVHD. The patient underwent HSCT from her HLA-matched sister and went into complete remission with 55% donor T (CD3) cells and 70% donor myeloid (CD33) cells 44. Remissions lasting 2 to 13 years have been reported in three RA patients who received myeloablative conditioning and allogeneic HSCT for concomitant hematologic malignant disease 45. Remissions lasting several years have also been seen in vasculitis due to Behçet's and Wegener's disease 3646.

In a recent retrospective analysis of the European EBMT (ProMISe) database, the outcome of 38 allogeneic HSCTs in 35 patients was analysed (T. Daikeler, T. Hügle, D. Farge, M. Andolina, F. Gualandi, H. Baldomero, C. Bocelli-Tyndall, M. Brune, J.H. Dalle, G. Ehninger, B.E. Gibson, B. Linder, B. Lioure, A.M. Marmont, S. Matthes-Martin, D. Nachbaur, P. Schuetz, A. Tyndall, J.M. van Laar, P. Veys, R. Saccardi, A. Gratwohl, submitted for publication). In that study, 55% of the patients showed a complete clinical response of their refractory disease and a total of 79% of patients responded at least partially. TRM at 2 years was 22.1%, in line with results in conventional hematologic diseases 40. The probability of survival at 2 years was 70%. TBI-containing conditioning was associated with a higher mortality and responses tended to be better when conditioning included cyclophosphamide.

Safety of allogeneic HSCT has improved by employing reduced-intensity regimens, graft manipulation like T-cell depletion, and new methods to prevent GVHD such as administration of mesenchymal stem cells or T-regulatory cells 47. In a recent prospective clinical trial, 56 patients with severe acute GVHD were treated with mesenchymal stem cells obtained from HLA-identical siblings or haploidentical or HLA-mismatched donors: 30 of 56 patients had a complete response, and 9 had a partial response. No side effects were reported. Interestingly, patients with a complete response had lower TRM compared with those with partial or no response and they had a higher overall survival 2 years after HSCT 48. Conditioning with alemtuzumab, a monoclonal anti-CD52 antibody targeting B cells, T cells, and DCs, is increasingly used for nonmyeloablative regimens. In another recent study, GVHD was significantly reduced in patients with aplastic anemia albeit at the expense of a higher rate of cytomegalovirus reactivation 49.

In autologous HSCT, conditioning with high-dose cyclophosphamide, ATG, and/or TBI deletes the majority of autoreactive effector cells of the host. Both adaptive and innate immune cells, including B and T lymphocytes, monocytes, NK cells, and DCs, are affected. This intensive immunosuppressive effect is considered to be the key effect of HSCT on AD, at least in the short term. The infusion of autologous HSCs serves to shorten aplasia but is probably not essential for the direct anti-autoimmune effect 38. In animal models, better response rates are seen in early inflammatory disease than in chronic longstanding AD 11. The latter may be due to a more important role of the stromal cell compartment, difficulties in disentangling disease activity from damage, or reduced sensitivity of the immune system to tolerance induction in longstanding disease 50.

Survival of autoreactive host lymphocytes with persistence of autoantibody titres after autologous HSCT has been described after both myeloablative and nonmyeloablative conditioning. Such cells may contribute to relapses. This implies that, in those patients without relapses, there must be factors that either inhibit proliferation of autoreactive clones or render the host anergic. These may include nonspecific effects of immunoablative therapy, such as post-transplant lymphopenia and reduced levels of pathogenic autoantibodies, and more specific effects, such as the induction of regulatory T cells. In the proteoglycan-induced arthritis mouse model, the initial improvement after autologous HSCT corresponded to an increase in CD4⁺CD25⁺ cells 51. Initially, these T-regulatory cells did not express FoxP3. Subsequent FoxP3 expression, however, was associated with a further stabilisation of AD. Similar results have been found in patients with juvenile chronic arthritis 52. After autologous HSCT, autoreactive T cells changed from a proinflammatory phenotype (mRNA interferon-gamma, T-bet high) before HSCT toward a tolerant phenotype (IL-10 and GATA-3 high). T-regulatory cells markedly increased after autologous HSCT. In the first period, they reconstitute through homeostatic clonal expansion; after several months, a thymic-dependent naïve CD4⁺CD25⁺ T-regulatory cell regeneration is seen 52. Interestingly, most of the relapses after autologous HSCT occurred within the first 9 months after HSCT, before the thymus-dependent recovery of naïve T-regulatory cells. Therefore, it is postulated that a well-functioning thymus is needed to generate a functionally active CD4⁺CD25⁺ population. Evidence for thymic reactivation after autologous HSCT has been obtained from studies showing increases in T-cell receptor excision circles and CD31⁺ T cells and normalisation of new T-cell receptor repertoires 5354.

The interaction of two different immune systems in one individual makes the pathogenetic mechanisms more complex in the allogeneic situation. When first performed in patients with AD, myeloablative chemo(radio)therapy was performed in conjunction with allogeneic HSCT, in analogy to hemato-logical diseases, in order to eradicate the host immune system 41. However, as shown in both animal models and several patients, nonmyelablative conditioning equally yielded stable mixed chimerism, resulted in less GVHD, and had the same efficacy in AD as myeloablative conditioning 1243.

After allogeneic HSCT, immune reconstitution depends on different factors such as stem cell source (peripheral blood versus bone marrow [BM], syngeneic versus allogeneic, and HLA-matched versus -mismatched), graft manipulation (mainly T-cell depletion), and age of the patient. Recovery of innate immunity (NK cells, monocytes, and granulocytes) occurs rapidly following transplantation whereas that of adaptive immunity is delayed and mostly incomplete 55. Especially after T-cell depletion of the donor graft, persistent low CD4⁺ T-cell numbers are observed 5657. Reconstitution of CD4⁺ T cells is more dependent on thymic function than CD8 cells. The grade of immunosuppression also depends on the type of graft source: unmanipulated peripheral blood stem cell (PBSC) grafts contain more T cells than BM grafts with a preserved CD4/CD8 ratio and a less activated immunophenotype 58. Recovery is significantly faster for CD4 cells, circulating monocytes, and NK cells after HSCT from peripheral blood versus BM 59. Cytokine profiles of PBSC-harvested T cells are polarised toward type II responses and these T cells exhibit a reduced ability to respond to allo- or autoantigens 58. BM, in contrast, contains more T-regulatory cells and mesenchymal stem cells than PBSCs. Both of them possess strong immunosuppressive effects, which may explain why more infections are observed after BM transplantation compared with PBSCs 60.

The role of T-regulatory cells after allogeneic HSCT has been analysed in chronic GVHD. The deficiency of T-regulatory cells due to consumption during acute GVHD probably contributes to chronic GVHD 61. On the other hand, in the absence of GVHD, an increase of the anti-inflammatory IL-10 is described which may promote the formation of T-regulatory cells. Apart from the increase of IL-10 serum levels, reduced levels of interferon, tumour necrosis factor-alpha, and IL-17 from stimulated T cells of mice with chronic inflammatory arthritis were described after allogeneic HSCT 17. More specifically, autoantigen-presenting cells are postulated to be attacked and eliminated by the allograft 62. Donor precursor T or NK cells are attracted by host antigen-presenting cells, presenting so-called host minor histocompability antigens. DCs, specialised in presenting antigens to donor effector cells, are mainly involved in this setting and probably also in autoantigen (cross)presentation in AD. In most tissues, recipient DCs are replaced after transplantation due to rapid turnover of BM-derived precursors 63. However, in a number of sites, notably the skin, recipient DCs may persist and even self-renew for many months after transplantation.

In collagen-induced arthritis, donor NK cells specifically eradicated autoantibody-producing plasma cells whereas other plasma cells remained unaffected 64. In lupus-prone mice, a significant decrease of ANAs was observed after allogeneic and, to a lesser extent, after syngeneic HSCT 12. The decrease of pathogenic autoantibodies and switch of subclass toward IgG1 may also be relevant 17. The total immunoglobulin levels after allogeneic HSCT decrease early after transplantation. Meanwhile, there are several reports of resolution of fibrosis after allogeneic HSCT. In this context, two patients who received allogeneic HSCT for scleroderma showed a resolution of collagenous deposits in the dermis 41. So far, however, the mechanism of this remains unclear.