Antithymocyte Globulin for Treatment of the Bone Marrow Failure Associated with Myelodysplastic Syndromes Free

Myelodysplastic syndromes are clonal bone marrow disorders that usually affect older adults. An estimated 20 000 new cases are documented annually in the United States (1, 2). The incidence seems to be rising, partly because older persons represent an increasing proportion of the population and partly because of improved diagnostic ability (3). Myelodysplastic syndromes are characterized by ineffective hematopoiesis that leads to varying degrees and combinations of anemia, neutropenia, and thrombocytopenia and by a tendency to evolve to acute leukemia that is refractory to treatment. Treatment of the bone marrow failure that accompanies myelodysplastic syndromes is unsatisfactory. It relies mainly on supportive care with red blood cell and platelet transfusions, antibiotics, and combinations of hematopoietic growth factors, which may partially improve blood cell counts. Although death from myelodysplastic syndromes can occur from leukemic progression, almost half of all deaths are due to cytopenia occurring without leukemic transformation (4). Therefore, improved treatments to relieve bone marrow failure associated with myelodysplastic syndromes should be greatly beneficial in terms of survival and quality of life.

Laboratory data suggest that immune mechanisms may contribute to the cytopenia of myelodysplastic syndromes (5-7). Case reports and small studies have shown that in some patients with the hypoplastic myelodysplastic syndrome, cytopenia responds to immunosuppressive treatment with antithymocyte globulin (ATG) or cyclosporine (8-10). In a previous report, we documented the remission of dependence on red blood cell transfusions and correction of neutropenia and thrombocytopenia in 12 of 25 patients with myelodysplastic syndromes who were treated with ATG (11).

In this study, we attempted to establish the frequency of red blood cell, neutrophil, and platelet response after ATG treatment in 61 patients (including the 25 reported in our earlier study [11]) with myelodysplastic syndromes. We also wanted to determine the survival and risk for leukemia progression in patients with myelodysplastic syndromes and to identify independent risk factors for treatment response, survival, and leukemic progression.

Between January 1994 and June 1998, patients were entered into a phase II trial of ATG treatment for myelodysplastic syndromes (study 95-H-0189). Of the 71 patients who were evaluated for the study, 61 met the eligibility criteria. All patients gave written informed consent, and the Institutional Review Board of the U.S. National Heart, Lung, and Blood Institute approved the study. Patients 18 years of age or older with a diagnosis of myelodysplastic syndrome and Eastern Cooperative Oncology Group (ECOG) performance status score of 2 or less were eligible for the study. Protocol entry was restricted to patients who were dependent on red blood cell transfusions, with or without concurrent neutropenia or thrombocytopenia. Transfusion dependence was defined as having had at least three separate transfusions of two or more units of red blood cells at intervals of 2 to 4 weeks to maintain the hemoglobin level (after exclusion of blood loss as a cause of anemia). To minimize the effect of previous treatments on ATG responses, patients were required to discontinue all other treatments capable of stimulating bone marrow (that is, growth factors, cyclosporine, steroids, and androgens) at least 1 month before entering the study. Myelodysplastic syndromes were confirmed by bone marrow aspirate and biopsy and were classified according to the French–American–British criteria as refractory anemia, refractory anemia with ringed sideroblasts, or refractory anemia with excess of blasts (12). Patients with chronic myelomonocytic leukemia and those with refractory anemia and 20% or more blasts in the bone marrow in transformation to leukemia (refractory anemia with excess blast cells in transformation) were excluded. Dysplasia affecting a minimum of two hematopoietic lineages was required for diagnosis. Marrow cellularity was assessed by surveying a 1- to 2-cm core biopsy specimen taken from the posterior iliac crest. Before treatment, bone marrow cytogenetic analysis was performed for all patients. These data were used to assign a score to each patient by using the International Prognostic Scoring System (IPSS) (1). The most common reasons for excluding patients from the study were dysplasia affecting only a single lineage or a bone marrow biopsy that was inadequate to render a diagnosis.

Before treatment, allergy to ATG was determined by intradermal injection of 0.1 mL of ATG at a concentration of 5 mg/mL. Antithymocyte globulin treatment was contraindicated if erythema larger than 5 mm in diameter, compared with the saline control, developed. Eligible patients were given ATG (Pharmacia, Peapack, New Jersey), 40 mg/kg of body weight daily for 4 days. Antithymocyte globulin was administered intravenously over 4 to 8 hours with oral prednisone, 1 mg/kg daily [minimum of 40 mg/d], for 10 days and then tapered by day 17. Clinic visits were scheduled at 3, 6, and 12 months and then annually. Bone marrow aspirate, biopsy specimens, and bone marrow samples for cytogenetic analysis were obtained at 6 and 12 months and then annually. Excluding transfusions and antibiotics, patients received no other treatment for myelodysplastic syndromes for at least 6 months. The main criterion for response was independence from red blood cell transfusion. This criterion was met if 1) the patient was independent from transfusion for a minimum of 6 weeks with a sustained increase in hemoglobin level, or 2) the patient maintained a stable hemoglobin level within 8 months of ATG treatment, as measured from the time the patient last received a transfusion until reinstitution of red blood cell transfusions. Standard criteria for red blood cell transfusion were followed. Transfusions were given for hemoglobin nadirs between 7 and 10 g/dL, depending on subjective tolerance of anemia in individual patients. Platelet and neutrophil counts that were the highest at 3 and 6 months of follow-up were used as the peak response. Thrombocytopenia was defined as a platelet count less than 150 × 10⁹cells/L, and neutropenia was defined as a neutrophil count less than 1.0 × 10⁹ cells/L. In patients with severe thrombocytopenia with a platelet count less than 20 × 10⁹ cells/L, a response was defined as a sustained increase to at least 25 × 10⁹ cells/L. In patients with severe neutropenia with a neutrophil count of 0.5 × 10⁹ cells/L, a response was defined as an increase to 1.0 × 10⁹ cells/L or greater. Morphologic characteristics of bone marrow, cellularity, and karyotype were compared before treatment, after 6 months, and then annually.

Results were analyzed on an intention-to-treat basis. Therefore, we evaluated all patients enrolled in the study and receiving ATG, regardless of whether they completed the course of ATG. Survival was censored in December 2001, with the exception of three bone marrow transplant recipients who were censored at the time of transplantation. Survival and time to progression for the entire cohort and response duration for the responders were evaluated by using the Kaplan–Meier method (13), with pointwise confidence intervals based on the variance formula of Greenwood (14). Within-patient differences before and after ATG treatment were evaluated by using the paired t-test for continuous variables and permutation tests for categorized variables. Prognostic factors for response were assessed by using logistic regression (15). The Fisher exact test was used to compute univariate P values for categorical predictors, and the likelihood ratio chi-square test with one degree of freedom was used for continuous predictors. Cox proportional-hazards regression was used to ascertain the effects of study entry characteristics and response on survival and disease progression (16). The effect of response on survival and disease progression was analyzed by modeling response as a time-varying covariate. With this approach, when a patient is “in response,” the model assumes that the relative risk for death or disease progression differs by a fixed amount compared with a similar patient who is not “in response.” Significance in Cox analyses was determined according to the score statistic. All P values are two sided; a P value less than 0.05 was considered statistically significant. Confidence intervals, determined by using Wald methods, were used for odds ratios. Stepwise regression methods using a modified step-up algorithm were used. At each step, a score statistic criterion (P = 0.25) was used for a variable to be entered into the model and a Wald statistic criterion (P = 0.30) was used for a variable to remain in the model.

These studies were conducted at the U.S. National Institutes of Health and were funded by the U.S. government. No outside agency influenced our analysis or interpretation of the data or the decision to submit the data for publication. Patient care was provided by the National Institutes of Health, and all drugs used in the study were purchased at full cost by the U.S. government.

Patient characteristics are shown in Table 1. Sixty-one patients were entered into the study. The median age was 60 years, and slightly more men than women were enrolled. The median duration of disease before ATG treatment was 18 months (range, 2 to 305 months). Forty-one patients (67%) had an IPSS score of intermediate-1, and 37 patients (61%) had refractory anemia subtype. In 46% of patients, karyotypic abnormalities, primarily involving chromosomes 5, 7, and 8, were identified. Before treatment, all patients were dependent on red blood cell transfusions (median duration of transfusion dependence, 7.5 months [range, 2 to 138 months]). Forty-one patients (67%) had platelet counts less than 150 × 10⁹ cells/L. Of these 41 patients, 21 (34%) had severe thrombocytopenia (≤ 20 × 10⁹ cells/L). Forty-one patients (67%) had neutrophil counts of 1.5 × 10⁹ cells/L or fewer. Of these 41, 11 (18%) had severe neutropenia (≤ 0.5 × 10⁹ cells/L). Six patients had severe pancytopenia.

Sixteen patients (26%) had received no treatment other than red blood cell transfusions. Before study entry, 21 patients (36%) had received some form of immunosuppressive treatment with corticosteroids, cyclosporine, or cyclophosphamide. Thirty-seven patients (60%) had been treated with androgens or hematopoietic growth factors, and 6 patients (10%) had received B vitamins. No patients had shown any substantial hematologic responses to any of these treatments.

All 61 patients entered into the study received ATG, but only 25% to 50% of the ATG dose was given in 3 patients because of unacceptable toxicity (fivefold increase in alanine aminotransferase and aspartate aminotransferase levels in 2 patients and cardiovascular instability in 1 patient). All patients developed some degree of serum sickness (consisting of fever, rash, and arthralgia) between 7 and 14 days after treatment. No other immediate or delayed toxicity resulted from ATG treatment. Fourteen patients progressed, six to refractory anemia with excess blasts in transformation and eight to acute myelogenous leukemia. Three patients with persistent pancytopenia received an allogeneic bone marrow transplant. Thirty-eight patients were censored alive, with an actuarial survival of 64% ± 13% at a median follow-up of 44 months (range, 1 to 88 months). Twenty-three patients died, 12 from cytopenia (3 from intracranial or pulmonary hemorrhage and 9 from bacterial or fungal sepsis) and 11 following progression to refractory anemia with excess blasts in transformation or to acute myelogenous leukemia (Table 2).

Twenty-one of 61 patients (34%) became transfusion independent within 8 months of ATG treatment. This group was defined as ATG responders (Table 2). Median time to transfusion independence was 10 weeks; 16 patients responded within 3 months. Transfusion independence was maintained in 17 of the 21 responders (80%), and the actuarial probability of remaining transfusion independent at 5 years was 76% (95% CI, 57% to 100%). Three additional patients who did not originally respond became transfusion independent at 18, 21, and 23 months after ATG treatment. Two of these 3 patients became tranfusion independent after treatment with erythropoietin; the third received only 25% of the ATG dose. Reticulocyte counts increased in responders but not in nonresponders; peak values coincided with transfusion independence. The mean reticulocyte count increased from 49 × 10⁹ cells/L ± 33 × 10⁹ cells/L to 85 × 10⁹ cells/L ± 44 × 10⁹ cells/L in responders (P < 0.001) and from 27 × 10⁹ cells/L ± 17 × 10⁹ cells/L to 30 × 10⁹ cells/L ± 30 × 10⁹ cells/L in nonresponders (P = >0.05). Three patients became transfusion dependent after remaining transfusion independent for 1.5 to 36 months. One of these patients retained a platelet response and regained independence from red blood cell transfusion after a second course of ATG, 30 months after the first treatment.

Of 21 patients with severe thrombocytopenia (platelet count ≤ 20 × 10⁹ cells/L), 16 (12 responders and 4 nonresponders) achieved sustained (untransfused) platelet counts between 25 × 10⁹ cells/L and 217 × 10⁹ cells/L. The mean platelet count increased significantly from 68 × 10⁹ cells/L to 137 × 10⁹ cells/L (P < 0.001) in responders and from 135 × 10⁹ cells/L to 141 × 10⁹ cells/L (P = >0.05) in nonresponders. Of 11 patients with severe neutropenia (neutrophil count ≤ 5 × 10⁹ cells/L), 5 of 6 responders and no nonresponders achieved sustained counts greater than or equal to 1.0 × 10⁹ cells/L. The mean neutrophil count increased significantly from 0.91 × 10⁹ cells/L to 1.95 × 10⁹ cells/L (P < 0.001) in responders and from 1.72 × 10⁹ cells/L to 1.73 × 10⁹ cells/L (P = >0.05) in nonresponders. The number of lineages affected by cytopenia improved significantly (P < 0.001) in responders: Before treatment, all 21 responders had neutropenia with or without thrombocytopenia and were transfusion dependent. After treatment, 19 of the 21 responders had normal neutrophil and platelet counts.

With the exception of one responding patient who progressed to acute myelogenous leukemia, no change was seen in dysplastic features of the bone marrow aspirate at 6-month follow-up examinations. No consistent difference was seen in bone marrow cellularity at 6 months when compared with the pretreatment biopsy appearance. Cytogenetic abnormalities that were present in four of the responders persisted after treatment. One patient developed a 7q− abnormality without loss of hematopoietic response.

We used logistic regression models to determine the effect of pretreatment patient characteristics on the probability of response to ATG (Table 3). Neutrophil and platelet counts were transformed by taking the base 10 logarithm. In the univariate analyses, the most significant factors related to favorable response were younger patient age, myelodysplastic syndromes subtype of refractory anemia, and lower platelet counts. Contributing to a lack of response were abnormal karyotype, anemia as the only cytopenia, hypercellular marrow, older patient age, and higher platelet count. In the multivariate model, each decade of life corresponded to an odds ratio for decreasing response of 0.45 (CI, 0.27 to 0.77); a 1-log increase in platelet count had an odds ratio for response of 0.34 (CI, 0.12 to 0.92).

The 5-year survival probability for the cohort of 61 patients was 64% (CI, 52% to 76%) (Figure 1, top); 23 patients (38%) died before the end of the study. In univariate analyses, the important characteristics that affected survival were age, type of myelodysplastic syndrome, neutrophil count, IPSS score, and treatment response (Table 4). However, only age, neutrophil count, and treatment response were selected by stepwise regression to be in the multivariate model (Table 4). Moreover, although IPSS score was strongly significant in the univariate analysis, its inclusion in the multivariate model did not significantly improve the model's fit (P = 0.15); the same was true for the type of myelodysplastic syndrome (P = 0.30). Figure 1 (bottom) shows the survival curves for the responders and nonresponders, with adjustment for age and neutrophil count. These curves branch at 8 months after treatment, because this is the time by which response was ascertained. Before the end of the study, only 1 of the 21 responders (5%) but 22 of the 40 nonresponders (55%) died.

For the 61 patients in the study, the 5-year probability of remaining progression free was 73% (CI, 61% to 85%) (Figure 2, top). Fourteen patients (22%) progressed during the study. In univariate Cox regression analyses, type of myelodysplastic syndrome, IPSS score, and treatment response were important predictors of progression (Table 5). Type of myelodysplastic syndrome, neutrophil count, and treatment response were chosen by stepwise regression to be in the multivariate model. Of these variables, only the type of myelodysplastic syndrome was significant in the multivariate model; treatment response was almost significant (Table 5). Moreover, although IPSS score was strongly significant in the univariate analysis, its inclusion in the multivariate model did not significantly improve the model's fit (P = 0.16). Figure 2 (bottom) shows the time-to-progression curves for the responders and nonresponders, with adjustment for type of myelodysplastic syndrome and neutrophil count. These curves branch at 8 months after treatment, because this is the time by which response was ascertained. Before the end of the study, only 1 of the 21 responders (5%) but 13 of the 40 nonresponders (33%) progressed.

We found that after ATG treatment, approximately one-third of the study patients with myelodysplastic syndromes achieved a sustained independence from red blood cell transfusion, with concomitant improvement in neutropenia and thrombocytopenia and prolonged survival without progression to leukemia. In univariate analyses, factors strongly predictive of response were younger age, refractory anemia subtype, and lower platelet counts. This analysis extends preliminary observations that ATG induced transfusion independence in some patients with myelodysplastic syndromes (8-11). Furthermore, although response was sustained in 17 of the 21 responders until end-of-study censoring, this does not take into account the possibility, seen in one of the patients who relapsed, of reestablishing transfusion independence with a second course of ATG. Three additional patients ultimately became transfusion independent after a long delay. However, these responses cannot be ascribed to ATG alone because these patients received other treatments, including erythropoietin, during follow-up.

The mechanism underlying the response to ATG is not clear. In contrast to aplastic anemia, in which marrow failure is due to an autoimmune, lymphocyte-mediated myelosuppression that is reversible by immunosuppression, marrow failure in myelodysplastic syndromes has been ascribed to ineffective hematopoiesis associated with the preleukemic state (17). However, laboratory studies suggest that immune-mediated myelosuppression also occurs in myelodysplastic syndromes. Autologous T lymphocytes from patients with myelodysplastic syndromes can suppress granulocyte and erythroid marrow cell progenitors. The suppressive activity is lost after ATG treatment (5-7). These in vitro findings and the clinical responses of myelodysplastic syndromes to immunosuppression with ATG or cyclosporine (8-11) suggest that the marrow failure in some patients with myelodysplastic syndromes is caused predominantly by an immune-mediated process similar to that described in aplastic anemia (17). Univariate analyses showed that younger patients with refractory anemia and lower platelet counts were more likely to respond. In a multivariate analysis, we found that marrow hypocellularity rather than a diagnosis of refractory anemia was an almost significant independent factor predicting response to ATG. However, the number of patients in our study is too small to confidently exclude the subtype of refractory anemia as a contributing factor for response.

Using Cox regression, we confirmed the previously reported effect of age and neutrophil count on survival and the predictive effect of neutrophil count and disease type on disease progression (1, 12). Moreover, we found that response to ATG was a strong independent variable favoring both survival and progression-free survival. The superior outcome for ATG responders was due to a low probability for leukemic progression and to the absence of cytopenia-related deaths. This was also apparent in the subset of 42 patients classified as intermediate-1 by the IPSS (1) (data not shown). Because both response and survival outcome were measured after ATG treatment, we cannot formally conclude that the response to ATG leads to prolonged survival. However, this is a likely conclusion because cytopenia improved and transfusions were no longer required in the responders.

Although our findings are of interest, they should be interpreted with caution. First, our study was not randomized, and without a control population of patients with matching prognostic criteria who received standard supportive care, it is not certain that treatment with ATG can confer an overall survival benefit in myelodysplastic syndromes. Furthermore, without a control group, the possibility that ATG may have unfavorably affected the outcome of nonresponders cannot be definitively excluded. The treatment did not appear to be detrimental, however, because the median survival of 5 months for patients with IPSS scores of either intermediate-2 or high was similar to the 6-month survival noted in the global analysis (816 patients), which was used to establish IPSS scores (1). We could not estimate the median survival of patients with either low or intermediate-1 IPSS scores because few deaths occurred in this small study; thus, no comparison could be made in these groups. Second, our study population represented a selected group of patients with myelodysplastic syndromes and severe cytopenia and anemia. The response to ATG would probably be lower in an unselected population of patients with myelodysplastic syndromes Finally, the statistical significance of response, when modeled as a time-varying covariate, is suggestive in conferring a survival benefit.

These findings raise the question of whether responders with hypoplastic bone marrow represented a subgroup of patients with a form of aplastic anemia. To avoid diagnostic confusion between myelodysplastic syndromes and aplastic anemia, we were careful to enroll into the study only patients with overt dysplasia in at least one lineage other than the erythroid series. However, accurate diagnostic separation of hypoplastic myelodysplastic syndrome from aplastic anemia is notoriously difficult and is limited by sample quality, observer differences, and the possibility of overlapping causes (18). Progenitor cell assays, which can help to determine the diagnosis in borderline cases (18), were not routinely performed in these patients. More recently, immunophenotyping using flow cytometry has been shown to help differentiate borderline cases (19). Finally, it should be noted that an assumption is made between cause and effect when ascribing delayed hematologic changes to the effect of a single 4-day course of ATG. Although it is unlikely, we cannot exclude the possibility that other treatments given before study entry could have affected the outcome.

In conclusion, these results have important implications for the management of myelodysplastic syndromes. They indicate that ATG should be further evaluated for the treatment of younger patients with myelodysplastic syndromes with refractory anemia subtype who are at risk from death from pancytopenia and in whom allogeneic stem-cell transplantation is not possible or indicated. Such patients have no curative options and have responded poorly to chemotherapy and other experimental treatments. The results also indicate that additional studies exploring different immunosuppressive regimens are warranted.