If a simple absence of stem cells were in fact the cause of aplastic anemia, the transfer of syngeneic bone marrow between identical twins who share the immunologic markers that define "self" should not require conditioning with cytotoxic drugs or irradiation to suppress the host's immune system. However, as shown in a literature review done in 1984 [3] and in the analysis of European registry data by Hinterberger and colleagues in this issue [4], most transplants from identical twins fail without such conditioning but succeed if the host's immune system is pharmacologically suppressed. Success with second transplantations that were preceded by immunosuppressive conditioning argues against a stromal-cell defect as the cause of initial graft failure. The absence of serious graft-versus-host disease suggests that minor degrees of histoincompatibility were not responsible for the high rate of graft rejection. Hinterberger and colleagues inferred that in most of their twin patients and, presumably, in most patients with aplastic anemia, underlying autoimmunity was the mechanism of bone marrow aplasia.

An immune basis for aplastic anemia was first suggested in the 1970s, when Mathe and colleagues [5] observed that recovery of autologous hematopoiesis (cells of patient origin) occurred after the transplantation of bone marrow from only partially matched donors had failed. Mathe and colleagues' patients had been prepared for transplantation with an immunosuppressive conditioning regimen that included antilymphocyte globulin. Antilymphocyte and antithymocyte globulin are immunoglobulin preparations made from the plasma of horses inoculated with human lymphocytes. The subsequent deliberate use of antilymphocyte globulin in Europe and, later, of antithymocyte globulin in the United States led to hematologic improvement in about half of patients with aplastic anemia.

As a result of these favorable clinical findings, basic laboratory investigations of aplastic anemia focused on the pathophysiologic role of the immune system [6]. Early experiments showed that patients' bone marrow or blood cells prevented normal cells from forming hematopoietic colonies and that removal of T cells from aplastic bone marrow improved colony formation in tissue culture. Activated cytotoxic lymphocytes were detected in the blood and bone marrow of many patients, and the numbers of these cells decreased after immunosuppressive therapy. The inhibitory lymphokines interferon- and tumor necrosis factor at least partly mediate the effects of lymphocytes on the proliferation of bone marrow cells. Aberrant expression of the interferon- gene was found more frequently in aplastic bone marrow than in blood, and even minute quantities of interferon that is locally produced in bone marrow potently suppress hematopoiesis [7]. Lymphokines can directly induce programmed cell death or apoptosis in hematopoietic cells [8]. The power of the immune system to destroy bone marrow is dramatically shown in the serious iatrogenic syndrome called transfusion-associated graft-versus-host disease, in which even small numbers of functional lymphocytes inadvertently inoculated into an immunocompromised host produce irreversible, fatal aplastic anemia. In this syndrome, the host's "foreign" HLA antigens are recognized by the transfused lymphocytes. The antigens in acquired aplastic anemia are not well defined, but aplastic anemia can occur after drug exposure, after such viral infections as hepatitis, and even during pregnancy. It is unclear how the antigens to which so many persons are exposed incite immune-mediated marrow destruction in a rare individual patient.

Antilymphocyte globulins are only mildly immunosuppressive, and their cytotoxicity for T cells produces a short period of moderate lymphopenia. Hematologic responses to antilymphocyte globulins are more often partial than complete but result in enough improvement to eliminate dependence on transfusion and to prevent infection. Complete normalization of blood counts occurs less frequently. In contrast to such biological reagents as antilymphocyte globulin and antithymocyte globulin, the drug cyclosporine has well-defined mechanisms of actions. Cyclosporine is not cytotoxic but does inhibit several lymphocyte functions, including the production of lymphokines and the proliferation of T cells. When cyclosporine was given as salvage therapy, about 50% of patients in whom antilymphocyte globulin or antithymocyte globulin had failed became transfusion independent. In a randomized multicenter trial in Germany [9] and in large studies from Europe [10] and the National Institutes of Health [11], the rational combination of antilymphocyte globulin and cyclosporine yielded hematologic response rates of 70% to 82% and 3- to 5-year survival rates of 64% to 92%. In contrast to treatment with antilymphocyte globulin or antithymocyte globulin alone, this intensive combination immunosuppressive therapy has been equally effective in children and adults and in patients with very severe neutropenia (<200 neutrophils/micro L).

Bone marrow transplantation is an historically established standard for patients with aplastic anemia, but is this arduous and expensive procedure always necessary? Immune-mediated disease of other tissues are usually treated first by drug therapy rather than by the initial physical replacement of the target organ! Clinical data relevant to this question derive from two sources: single-center studies and transplantation registries. Reports from single institutions may favor either transplantation [12, 13] or immunosuppression [14], but overall survival figures in such studies are not statistically significantly different, perhaps because of small numbers of patients. Data from European registries on very large samples have consistently shown that the two treatments are associated with equivalent long-term survival rates ([15, 16]; Bacigalupo A. Personal communication). Single-center studies may highlight the superiority of specific protocols, but registry data may be less prone to various biases because of different referral patterns to individual tertiary centers; rivalry among centers in increasingly competitive economic and academic markets; and the effect of patients' medical, social, and, especially, financial circumstances on treatment choices and outcomes.

In a retrospective analysis published in this issue, Doney and colleagues [17] conclude that transplantation is superior to immunosuppression as first treatment for aplastic anemia, at least in patients younger than 40 years of age. However, as a guide to current practice, this study suffers because of the outdated regimens used in the immunosuppression group. Many patients received antithymocyte globulin in combination with haploidentical bone marrow infusions, androgens, or high-dose methylprednisolone, none of which has been shown to improve responses. Conversely, only one patient in Doney and colleagues' immunosuppression group received cyclosporine. The importance of cyclosporine in the treatment of aplastic anemia is reflected in retrospective reviews, which show that a critical variable in the prognosis of aplastic anemia is the period during which patients are treated. As noted, cyclosporine (which was introduced for the prevention of graft-versus-host disease in the mid-1980s) dramatically improved the outcome of bone marrow transplantation. Since the early 1990s, cyclosporine combined with antilymphocyte globulin or antithymocyte globulin has greatly increased hematologic response rates in immunosuppression protocols [12, 16]. Thus, the study by Doney and colleagues [17] compares bone marrow transplant recipients, most of whom received cyclosporine as prophylaxis, with patients who received antithymocyte globulin without the additional benefit of cyclosporine.

By necessity, immunosuppression is the treatment of choice for most patients. Very few persons must choose between transplantation and immunosuppression. A calculation based on the incidence of aplastic anemia (2 cases per 1 million persons), the age distribution of patients (median, 25 years), and the 25% probability of having a fully histocompatible sibling yields a Figure of fewer than 100 such patients per year in the United States. Experts agree that children with sibling donors do extremely well after transplantation; for adults, however, each therapy has advantages and disadvantages. Transplantation cures bone marrow failure, but sometimes at a cost of severe complications, such as graft-versus-host disease, graft rejection, and infection. Intensive immunosuppression causes relatively mild side effects, but the underlying bone marrow process may be incompletely treated. About 30% of patients have relapse [18], and about 16% develop late secondary clonal hematologic diseases (especially myelodysplasia and leukemia), sometimes years after apparent bone marrow recovery [19].

Because a modest transfusion burden has not diminished success rates for transplantation [20], some patients may reasonably delay this procedure and opt first for a trial of immunosuppression; transplantation could be deferred to treat refractory disease or the later development of myelodysplasia [21]. More potent immunosuppressive agents, such as cyclophosphamide [22] and purine analogues, alone or in combination with cyclosporine, are currently being studied. Improvement in both transplantation and immunosuppression will continue to benefit patients with aplastic anemia. Finally, and somewhat ironically, despite the availability of two excellent treatments, patients often still receive ill-considered and sometimes harmful trials of corticosteroids or growth factors rather than prompt referral for transplantation or immunosuppression [16].