Chapter 123:  Myelodysplastic Syndrome

Clonal Origin

Since originally suggested by Dacie, a substantial body of evidence has accumulated that points to a clonal origin for myelodysplastic syndromes. The identification of nonrandom chromosomal abnormalities detectable in 30 to 70% of patients with MDS confirmed the likelihood that these were clonal disorders.^21,62–67 Evidence that these disorders originate from pluripotent stem cells was suggested by reports of patients developing biphenotypic and lymphoid leukemias following MDS.^68–72

Analysis of the patterns of inactivation of an X-linked polymorphic enzyme, such as glucose-6-phosphate dehydrogenase (G6PD), has proven to be a useful tool in the analysis of the clonal origin of tumors.⁷³ Raskind and colleagues have demonstrated that 21 of 24 B lymphoblastoid lines transformed by Epstein-Barr virus expressed a single light chain immunoglobulin and contained only one of the isoenzymes. In contrast, the T-cells expressed both the A and B G6PD isoenzymes.⁷⁴ The frequency with which the lymphoid lineage is involved as part of the abnormal clone is quite variable as demonstrated in a number of recent studies.^75–79

Subsequent studies using more refined and sensitive molecular techniques have also suggested that cells of lymphocytic origin belong to the abnormal clone, with B lymphocytes being involved more regularly than T cells.^75–78 However, a number of other studies have found conflicting evidence suggesting that only cells committed to the myeloid lineage are involved in the clone, but that lymphocytes are of polyclonal origin.^75,77,79 Studies using fluorescent in situ hybridization (FISH) or polymerase chain reaction (PCR) analysis of loss of heterozygosity have confirmed the involvement of an early stem cell which can give rise to CD34+ cells as well as those of erythroid, megakaryocytic, and myelomonocytic lineage but not lymphocytes.⁸⁰ The reason for the discrepant findings is unclear but may reflect the heterogeneity of the disease and interpatient variability. A recent study supports this concept where the bone marrow from 2 of 5 patients with MDS manifests evidence of residual polyclonal hematopoiesis.⁸¹ Thus, the diverse heterogeneity of disease biology at the clinical level may simply reflect the degree of heterogeneity of the disease at the cellular level.

In Vitro Progenitor Growth Characteristics

A variety of abnormalities in the growth of hematopoietic progenitors has been detected in colony assays in vitro of both bone marrow and peripheral blood from patients with all categories of MDS.^82–87 In a series of studies, 74% of 170 patients had abnormal bone marrow myeloid progenitor growth (CFU-GM). The pattern of colony growth can be classified as either a leukemic or a nonleukemic pattern.^83,85,86 The former are characterized by increased formation of micro- and macroclusters; defective maturation of the colonies formed with the persistence of blasts; or decreased or absent colony formation (< 2 colonies/10⁵ bone marrow cells plated). A leukemic type growth pattern is more commonly associated with shortened survival and an increased risk of transformation to acute leukemia than a nonleukemic pattern.^85,86 In patients serially studied, progressive increases in the cluster/colony ratio or progressive decrements in the growth of CFU-GM were seen prior to or concurrent with the transformation to leukemia. A positive correlation between abnormal colony growth and the presence of an abnormal karyotype has been identified and is associated with an increased risk of leukemic transformation. Although less extensively studied, reports have also indicated decreased to absent growth for other progenitors including those of erythroid (CFU-E, BFU-E),^75,88,89 mixed lineage (CFU-GEMM),^90–92 and megakaryocytic (CFU-Mk)^93,94 precursors. No consistent pattern of colony growth in association with FAB subtype has been demonstrated, but some studies have indicated the appearance of increasing growth abnormalities in relation to increasing numbers of bone marrow blasts.⁹³ CMML is often distinguished by increased CFU-GM growth, even in the absence of exogenous growth factor supplementation.⁹⁵

Production of endogenous regulatory factors influencing the growth of myeloid, erythroid, and megakaryocytic precursors is diminished.^92,94,96 Bone marrow conditioned medium from patients with MDS had both decreased CFU-GEMM_CSA and burst promoting activity (BPA), compared with conditioned medium from normal marrow. Both these activities were completely neutralized by anti-interleukin (IL)-3 antibodies.⁹² In addition, accessory cells, including macrophages and natural killer (NK) cells, appear to be capable of directly or indirectly inhibiting bone marrow CFU-GM through the production of soluble factors.^88,97 T cells, although abnormal in number and in their response to mitogens, appear to produce normal levels of growth factors.⁹⁸ Partial restoration of colony growth can be accomplished through the addition of exogenous growth factors or other agents to the culture.⁸⁹ The addition of rhGM-CSF can result in an increase in CFU-GM but has little effect on the growth of CFU-GEMM, BFU-E, or CFU-Mk.^87,90 Colony growth of CFU-E and BFU-E in some subgroups of MDS improved with the addition of rhEpo (recombinant human erythropoietin) or kit ligand (steel factor).⁹⁹ Addition of rhG-CSF (recombinant human granulocyte colony-stimulating factor) in vitro partially restored the abnormal function of neutrophils derived from patients with MDS.¹⁰⁰ These observations may be of clinical importance.

Studies of cytokine production by the bone marrow stroma have produced variable results. Stromal cell production of a variety of cytokines including GM-CSF, IL-6, IL-1ß, tumor necrosis factor (TNF)-α, and IL-8, has been reported to be impaired in some patients with MDS and can be restored, at least in part, by in vitro or in vivo administration of hematopoietic growth factors.^101,102 However, the level of G-CSF, IL-1β, IL-6, IL-8 or stem cell factor messenger RNA (mRNA) in a small series has been reported to be normal or increased.¹⁰³ Serum cytokine levels have, for the most part, been found to be normal or increased except for stem cell factor, which in one series was decreased.^104–107

Bone Marrow Microenvironment

Histologic examination of bone marrow trephine biopsies by Tricot and colleagues have pointed to abnormalities of the microenvironment.¹⁰⁸ They noted the presence of clusters of immature precursor cells in the central intertrabecular region of the marrow, rather than along the endosteal surfaces. They cited this as evidence of abnormal localization of immature precursors (ALIP).¹⁰⁸ ALIP was detected even before bone marrow smears revealed an excess of blasts. In a series of 40 patients, the presence of ALIP correlated significantly with shortened survival and was associated with an increased risk of transformation to AML. These findings were independent of the FAB subtype and were detected even in patients with refractory anemia. Care must be applied to differentiate true ALIP from pseudo-ALIP. In the latter case, the clusters of cells are either of erythroid or megakaryocytic origin and do not convey the same prognostic information, compared with the former, where the immature cells are of myeloid origin. The determination of the immature precursor phenotype by immunohistochemical methods may be helpful in distinguishing pseudo-and true ALIP and thus permit identification of specific subgroups with a poor prognosis.¹⁰⁹ ALIP is not, however, specific to patients with MDS and, therefore, not helpful as a diagnostic tool.¹⁰⁹

Investigations to define the potential role the bone marrow stroma may play in MDS have been limited and have yielded conflicting data to date. Some studies, however, have suggested that abnormalities of the stroma contribute to the pathophysiology of MDS. In vitro stromal cells either fail to reach confluent growth or their growth is absent in the majority of patients. Furthermore, stromal cell support of normal hematopoiesis in long-term bone marrow cultures can be impaired.^110–112 Cytokine production of (LIF) and TNF-α by MDS stroma is also impaired. In vivo labeling techniques have demonstrated that the ineffective hematopoiesis in MDS is accompanied by a high cell turnover, with an increase in apoptosis of both hematopoietic progenitors and stromal cells. In vitro analyses indicate that the areas of apoptosis are associated with high levels of TNF-α and low levels of GM-CSF.^111,113,114

Signal Transduction

Whether the primary pathophysiologic defect(s) resulting in disordered maturation and function are intrinsic to hematopoietic progenitor cells or derive from their interaction with accessory cells and other microenvironmental (ME) factors or a combination thereof is uncertain. Hematopoietic cell defects may relate not only to quantitative abnormalities of progenitors but also reflect abnormalities in signal transduction in response to cytokines or other regulatory molecules that trigger proliferation and differentiation pathways. Hematopoietic cells from MDS patients display impaired responses to a number of cytokines and are unable to respond to external signals due to either abnormalities in number or function of cytokine receptors or to dysfunctional postbinding signal transduction.⁹⁶

Hematopoietic progenitors have impaired responses to cytokine stimulation with decreased CFU-GM, CFU-GEMM, and BFU-E colony number.^89,98,115 Purified blast cells from MDS patients proliferate but do not mature in response to G- and GM-CSF. Purified populations of CD34+ cells from MDS patients also demonstrate impaired responses to G-CSF.¹¹⁶ Receptor abnormalities in MDS are not commonly observed,^117–120 although in some patients, their number is either diminished or their structure aberrant.¹²¹ Progenitor differentiation to erythropoietin is blocked, and cells undergo apoptosis in patients with a truncated erythropoietin receptor.¹²¹ On the other hand, defects in signal transduction appear to play a larger role in the aberrant response of hematopoietic progenitors to regulatory molecules.^122,123 This pertains both to the cytokine signal transduction pathways and factors which regulate apoptosis.^123,124 Patients with MDS have defective activation of STAT5 in response to erythropoietin but not IL-3, suggesting a defect in the erythropoietin signaling pathway. Despite the relative impairment, increasing concentrations of cytokines can partially correct the defects.^125,126

The majority of patients with MDS have hypercellular bone marrows with evident proliferation of various cellular lineage components. One interpretation of this finding is a compensatory response in the marrow to feedback signals secondary to the peripheral blood cytopenias. Yet, the response does not translate into effective hematopoiesis; a role for increased apoptosis has been suggested.^{124,127–129} A deficiency in cytokines or relative resistance to their effects could explain this phenomenon of ineffective hematopoiesis. In several recent studies, bone marrow cells from MDS patients have demonstrated increased expression of Fas and Fas-ligand. In marrow cultures, strategies that block TNF-mediated signals, such as the use of anti–TNF-α antibody, significantly increased the numbers of hematopoietic colonies, compared with untreated cells.¹²⁸ Additional studies have implicated a dysregulated Fas pathway and TNF-α as contributing to ineffective hematopoiesis.^127,129,130 The role of cytokine signaling with apoptosis is less well delineated but may contribute.^123,124 Progenitors with impaired signal transduction, thus constituting a refractory target cell, could undergo accelerated apoptosis analogous to withdrawal of obligate survival factors. Alternatively, the apoptotic pathway itself may be dysregulated with direct activation of the Fas pathway.¹²⁴ As MDS evolves to AML, the acceleration of apoptosis declines, and AML is characterized by increasing progenitor survival.¹³¹

Cytogenetics

Chromosome analysis contributed to the initial determination that MDS is a clonal disorder. In an analysis of studies on a total of 898 patients, 49% were identified as having an abnormal karyotype.^{21,27,66,67,132–134} With recent improvements in both banding and high-resolution synchronization techniques, 79% of patients in one study had abnormal karyotypes.⁶⁷ The chromosomal abnormalities are similar to those seen in patients with AML and involve complete or partial deletions, most frequently involving chromosomes 5(-5,5q-), 7(-7,7q-) and 8(8+) (Table 123.3). Certain karyotypic abnormalities that have been associated with specific morphologic leukemic subtypes in AML, such as t(15;17) in acute promyelocytic leukemia (M3), t(8;21) in acute myelocytic leukemia (M2) and t(9;11) in acute monocytic leukemia (M5), have only rarely been identified in MDS (75, 145–147) . This suggests that these particular leukemic subtypes are not preceded by a preleukemic phase. Furthermore, abnormalities involving partial deletions of the long arm of chromosome 20 (20q-), seen frequently in MDS, particularly RARS, polycythemia vera, and myeloproliferative syndromes, are usually not seen in patients with de novo AML.²¹ Abnormalities are identified most frequently in patients with RAEB and RAEB-T, compared with those with RA and RARS.^134,135 Although specific chromosomal abnormalities have been identified, unlike AML, no specific abnormality has been associated with any specific FAB category.^21,67,134

Many studies have demonstrated that the presence of karyotypic abnormalities is an independent prognostic variable.^21,27,67,134 Several studies have demonstrated that patients with abnormalities present in 100% of the mitoses (AA) examined, or a mix of normal and abnormal mitoses (AN) have poorer survival than patients with normal karyotypes (NN).^{21,66,67,135,136} Furthermore, more complex abnormalities are associated with shortened survival, compared with either single clonal abnormalities or a normal karyotype.^{27,67,133–135} Certain specific clonal subtypes have varying prognostic significance for both survival and the risk of leukemic transformation. Monosomy 7 or 5 and 7q- are associated with shortened survival, while deletions of the long arms of chromosome 20(20q-) and 5(5q- syndrome) and deletion of y(-y) as the sole abnormality are associated with prolonged survival.^{27,67,133,135,136} In the IPSS, cytogenetics findings are an independent prognostic variable classifying patients according to three subgroups: good, intermediate, and poor risk.²⁷

Genetic instability as demonstrated by clonal evolution occurs in a substantial percentage of patients. From 20 to 35% have been found to undergo clonal evolution during the course of the disease, independent of karyotypic status at the outset.⁶⁶ The significance of clonal evolution with respect to leukemic transformation and survival is contradictory at the outset.^{66,134,136,137} In some studies, clonal evolution was associated with a poor prognosis without an increased risk of leukemic transformation.¹³⁶ In contrast, Glenn and colleagues have demonstrated karyotypic changes in clinically stable patients.¹³⁷

One particular abnormality involving a deletion of part of the long arm of chromosome 5(5q-) deserves special note. As originally described, the 5q- syndrome is associated with a refractory macrocytic anemia, a normal or increased platelet count, giant thrombocytes, dyserythropoiesis, hypolobulated megakaryocytes, female predominance, prolonged survival, and a low rate of leukemic transformation.^{67,135,138,139} It is important to differentiate the 5q- syndrome from instances where this deletion is found in combination with other chromosomal abnormalities or in association with FAB subtypes other than refractory anemia. It is clear that those patients with the 5q- abnormality, without the characteristic clinical and morphologic features, have a more aggressive clinical course associated with short survival.

Patients with myelodysplastic syndromes secondary to exposure to mutagenic or carcinogenic agents have similar findings in terms of the types and significance of the chromosomal abnormalities.^62,140,141 Patients treated with epidophyllotoxins (etoposide) can develop specific translocations involving the break point at 11q23.^46,48 This leads to transcription of a fusion protein involving the mixed lineage leukemia (MLL) gene.^142,143

Oncogenes

Alterations in the control and expression of proto-oncogenes in the cellular genome, leading to abnormalities of cellular proliferation and differentiation are thought to contribute to the molecular basis of neoplastic transformation.¹⁴⁴ Point mutations of the ras family of oncogenes have been identified in association with a number of human tumors, including lung, pancreatic, colorectal, and hematopoietic neoplasms.^145–148 Activated ras genes with specific point mutations involving codons 12, 13, and 61 have been identified in 20 to 30% of patients with AML and 9 to 48% of patients with myelodysplastic syndromes. These findings have suggested that activation of ras genes may either contribute to the development of MDS or, once established, to the process of transformation to AML.^{147,149–152}

Mutations of all three ras genes have been demonstrated in 75 of 628 (12%) patients with MDS. (Table 123.4)^{147,149,153–158} N-ras was most commonly affected, followed by Ki-ras, with H-ras involved in only a few cases. CMML was the FAB subtype most frequently associated with a ras mutation. The abnormality was found in 40% of cases.¹⁵⁹ The association may, however, be a function of the monocytosis, since other FAB subcategories with ras mutations were often associated with increased monocyte counts.¹⁵⁸

In the initial reports, the presence of a mutant ras gene was associated with transformation of MDS to AML.^149,153 Overall 47% (65 of 137) of patients with a ras mutation transformed to AML, compared with only 14% (89 of 628) of those without a mutation.^{147,149–162} A significantly shortened survival of patients with MDS which transformed to AML has been associated with the presence of a ras mutation in most.^67,151 However, this has not been a consistent finding.¹⁶³

The ras mutation may confer a selective growth advantage to cells and appears to occur after initiation of leukemogenesis.^154,156,160 These mutations can appear in either the early or late stages of MDS (Fig. 123.1). In the early stages, the mutation is present in all bone marrow and peripheral blood cells, and involves a totipotent stem cell.^156,158 In other cases, it appears later in the disease and tends to be associated with an evolving and expanding clone of cells.^136,150,154 Its appearance in association with evolving karyotypic changes indicates an underlying genetic instability and suggests that ras mutations are not the last step in the transformation pathophysiology.¹⁵² It is also unclear whether ras mutations contribute to the transformation from MDS to AML, or merely represent a genetic epiphenomenon in the emergence of a new clone.

Recent data provide further insight into a potential role for ras mutations. CD34+ progenitors modified to express a mutated N-ras were found to have defective differentiation in response to erythropoietin. This was manifested as reduced proliferation, increased doubling time, and decreased number of cells in S/G2M, leading to a failure to express the differentiation program in the late erythroblast stage. These cells also tended to undergo accelerated apoptosis, suggesting that expression of the mutated N-ras may be responsible, at least in part, for the pathophysiology underlying MDS.¹⁶⁴

Deletions of all or part of chromosome 5 in a substantial number of patients with either de novo or secondary MDS has led to the hypothesis of a tumor suppressor gene located within a short segment of the long arm in the so-called critical region at 5q21 - 5q34.¹⁶⁵ The interferon regulatory factor-1 (IRF-1) is a DNA-binding regulatory factor, which regulates, in part, expression of interferon and interferon-inducible genes by binding to promoter regions. It functions as an activator, has antiproliferative and antioncogene activity, and can reverse a transformed phenotype.¹⁶⁶ Willman and colleagues have mapped the IRF-1 gene to the 5q31 region.¹⁶⁷ They identified a loss of one or both IRF-1 alleles in all of 13 patients with MDS (6) or AML with a del 5q.¹⁶⁷ In two other studies, an additional 15 of 19 patients with MDS and a del 5q were also found to have a deletion of one or both alleles.^168,169 In an elegant study using embryonic fibroblasts from knockout mice, Tanaka and colleagues demonstrated that a null mutation in the IRF-1 gene when combined with expression of the Ha-ras oncogene results in a transformed phenotype.¹⁷⁰ However, expression of Ha-ras with either the wild type or IRF-2 knockout is not associated with transformation. Furthermore, the loss of IRF-1 alleles rendered cells more resistant to chemo- or radiation-induced apoptosis. These data suggest a potential role for the IRF-1 gene in MDS and warrant further study.

Point mutations in the p53 tumor suppressor gene have been reported in 8% of MDS cases.^{171, 172} In one series, patients with secondary MDS (sMDS) or therapy-related AML (tAML) were found to have a higher than expected prevalence of p53 mutations in leukemic, but not germline, tissues. Microsatellite instability was identified and was consistent with a mutator phenotype, suggesting that these patients were at higher risk for developing therapy-related MDS/AML.¹⁷³ Recently, patients have been described with a 17p- syndrome characterized by dysgranulopoiesis with pseudo-Pelger-Huet hypolobulation and small vacuoles in neutrophils. In one series, 15 of 16 patients with this deletion had an associated deletion of p53, suggesting a potential role for loss of a tumor suppressor gene as a contributing factor to the morphologic, cytogenetic, and molecular phenotype of this syndrome.¹⁷⁴ The expression of several other genes including EVI-1, c-mpl, PDGF, MLL (mixed lineage leukemia), and the CSF-1 receptor have recently been described to be dysregulated in some patients with MDS.^{143,168,175,176} The EVI-1 gene located on chromosome 3 has been shown to play a role in both myeloid and erythroid differentiation. Dysregulation of gene expression is associated with blocks in myeloid and erythroid maturation.^177,178

Hormones

Although utilized as a nonspecific stimulant of erythropoiesis in aplastic anemia with some benefit, androgens have not proven to be useful in MDS. Initial observations suggesting that treatment with androgens accelerated the disease and was associated with an increased rate of leukemic transformation was not confirmed in subsequent studies, nor was any efficacy identified.^215,216

Glucocorticoid therapy yielded an overall response rate of only 9%, and 24% of nonresponding patients experienced significant deleterious side effects.²¹⁷

Danazol has been reported to produce improvements in the thrombocytopenia and anemia in patients with MDS. The effects appear to be mediated, in part, through inhibition of Fc receptor expression on monocytes, which alter the clearance of immunoglobulin-coated platelets and red cells from circulation, rather than a direct stimulatory effect on hematopoiesis. Danazol may be helpful in those patients in whom immune-mediated cell destruction contributes to the deficit.²¹⁸

Chemotherapy

Chemotherapeutic agents, alone and in combination, have been utilized to treat patients with MDS. These agents have been employed in a variety of regimens ranging from attenuated low-dose schedules⁸⁶ to the more conventional antileukemic myelotoxic type strategies. These have been employed to treat patients at all stages of disease. Antileukemic type treatments have not substantially altered the outcome of the disease for most patients and have been associated with significant toxicity in many.

The pharmacologic interaction between fludarabine and cytosine arabinoside increases intracellular ara-CTP and has been exploited in a regimen with activity in patients with relapsed AML. Subsequently, the combination with (FLAG) or without (FA) G-CSF has been tested in patients with MDS and de novo AML.²¹⁹ The two regimens yielded comparable complete response rates of 60% and 55%, respectively, with the response rate generally higher in the subgroups of MDS with excess blasts. Overall, 27% of patients (MDS and AML) died during induction. The median projected survival was 29 weeks for FA and 39 weeks for FLAG. These differences were not statistically significant. In patients with deletion of chromosome 5 or 7, FLAG produced a response rate of 64% versus 36% for FA. The differences were attributed by the authors to factors other than an effect of treatment, however. Use of G-CSF was associated with a more rapid recovery of neutrophil count but this did not translate into decreased rates of infection or infection-related mortality.²¹⁹ Overall, there were no differences in treatment outcome for patients with MDS, compared with those with AML.

In a successor study, 215 patients (153 AML and 62 RAEB/RAEB-T) were treated with fludarabine (F), cytarabine (A) and idarubicin (I), with or without either G-CSF or all-trans-retinoic acid (ATRA) (FAI; FAI + G-CSF; FAI + ATRA; FAI + G-CSF +ATRA).²²⁰ The response was 51%, and the median survival was 28 weeks. Addition of G-CSF +/– ATRA to FAI did not affect overall outcome.

In general, results from earlier studies indicated that responses to treatment of patients with either MDS or AML following MDS are less favorable than treatment for de novo AML.^221–223 Age appears to influence the rate and duration of response, with younger patients achieving complete response (CR) more frequently²²⁴ and remaining in remission longer,^222,225 compared with older patients. Achievement of a CR is associated with improved survival in comparison with nonresponding patients,^222,223 but the duration of the response is substantially shorter in comparison with patients with de novo AML who achieve a response.^221–223 Treatment prior to the transformation to AML,²²¹ and patients with shorter intervals between the diagnosis of MDS and leukemic transformation (280, 285) are associated with higher response rates. Aggressive antileukemic type treatment is associated with high rates of morbidity and mortality, however, with up to 30% of the patients dying from drug-related complications.^221–223 Recent studies at M.D. Anderson Cancer Center suggested that there were no differences in treatment outcome for their patients with MDS, compared with AML, and that diagnosis (RAEB, RAEB-T versus AML) was not predictive for outcome.

The topoisomerase I inhibitor topotecan has demonstrated activity in early phase II trials. As a single agent, topotecan produced a CR rate of 32% in patients with MDS, including a high proportion of patients with CMML.²²⁶ The treatment-related mortality rate was 20%, with a median survival of 45 weeks and 38% alive at 12 months. In a subsequent study, topotecan 1.5 mg/m²/d continuous intravenous infusion for 5 days was combined with cytarabine 1 g/m² daily for 5 days yielding a 47% early CR rate (by day 50) and a mortality of 2% when administered to hospitalized patients in a protected environment.²²⁷ Overall, survival was the same as with topotecan as a single agent. In fact, these survival data appear little better than the results for patients treated with FA, FLAG, FAI, FAI+G-CSF, FAI+ATRA, or FAI+ATRA+G-CSF. Thus, although response rates appear better than historical experience, this antileukemic strategic approach does not appear to translate to an effect on overall survival in changing the natural history of the disease.

The relative resistance of disease in patients with MDS and the pattern of emergence of resistance suggest that multi-drug resistance (MDR) may be a factor in the poor response rates in patients with MDS or AML following MDS. In one study, patients with MDS were found to have an increased expression of MDR mRNA. The MDR phenotype has been correlated with expression of the human progenitor cell antigen CD34. This stem cell phenotype, in association with MDR expression, may account for increased drug resistance.²²⁸

Bone Marrow Transplantation

Results of allogeneic and syngeneic bone marrow transplants from a series of reports containing small numbers of MDS patients have suggested that 35 to 40% can achieve durable long-term disease-free remissions when treated with this modality.^229,230 This has been substantiated with the results from two larger single-institution series.^231,232 In a recent update, 251 patients treated between 1981 and 1996 were evaluated.²²⁹ Appelbaum and colleagues reported an estimated (Kaplan-Meier) 5-year disease-free survival (DFS) of 40%. Younger age, shorter duration of disease, female gender, and de novo MDS were predictors of better DFS. Patients under the age of 20 years had a 60% disease-free survival rate, compared with 40% and 20% for those 20 to 50 years old or over the age of 50 years, respectively. Patients with low-risk MDS had a 55% DFS at 6 years, compared with only 30% for those with high risk. This difference correlated with a higher rate of relapse among those with more advanced disease. Among patients with low- and intermediate-1-risk groups, according to the IPSS score, there were almost no relapses. The 5-year DFS for those in the low and intermediate-1 groups was 60%, 36% for those in intermediate-2, and 28% for those in the high-risk group. Patients undergoing matched unrelated donor (MUD) marrow transplantations fare less well. In an analysis of results from the National Marrow Donor Program, patients receiving MUD transplants during the first 4 years of Registry data (1986–1990) had a disappointing DFS of only 18% at 2 years and an overall survival at 2 years of 24%.²³³ In one study, patients with primary MDS demonstrated a survival advantage (56%) over those with secondary MDS (27%).²³⁴

The exact role and timing of bone marrow transplantation remains to be determined, as does the optimal conditioning regimen. For patients 40 years or younger with a compatible related donor, however, transplantation should be favored, since no other therapy, thus far, is curative. The data demonstrate that particularly those MDS patients without excess blasts whose disease is of short duration should be strongly considered for marrow transplantation even though their relative longevity without transplantation is better than other MDS patients. Fewer patients over the age of 40 years have been transplanted, but limited data for those with related donors suggest that up to 30% may survive disease free.²³⁴ Transplantation for those without a related donor is a more difficult choice in view of the data noted above. Given the age of most patients with MDS and the availability of donors, transplantation in the foreseeable future is likely to be of limited value, benefiting only about 5% of all patients with MDS.

The use of high-dose chemotherapy with either bone marrow or peripheral blood stem cell (PBSC) infusion has been utilized in a limited fashion. The European Group reported on 79 patients treated with intensive chemotherapy and autologous bone marrow transplantation in first CR. Fifty-five of the 79 in whom the duration of first CR was known were matched with 110 patients with de novo AML. The 2-year survival for all 79 patients was 39%, and the DFS at 2 years was 28% for the 55 patients with MDS/sAML versus 51% for patients with de novo AML. Relapse rates were 69% for MDS/sAML and 40% for de novo AML (p = .007).²³⁵ Use of PBSC infusion is more problematic, given that adequate collection from patients with MDS is obtained in only half the patients.²³⁶

Differentiation Inducing and Novel Acting Agents

There has been great interest in the potential of differentiation therapy as an antitumor modality since Charlotte Friend and her colleagues first demonstrated that dimethylsulfoxide (DMSO) could induce differentiation of murine erythroleukemic cells in vitro, thus altering their malignant phenotype.²³⁷ The concept, while readily documented in vitro, is more difficult to demonstrate in the clinical setting. A number of agents which have effectively induced differentiation in vitro have been tested in MDS without significant success (e.g., cis- and trans-retinoic acid, vitamin D₃, butyrate, hexamethylene bisacetamide [HMBA]).

Cytosine arabinoside has been extensively tested. In an extensive review of the literature, low-dose cytosine arabinoside was found to produce 16% CR in 170 patients reported.²³⁸ Median duration of CR was 10.5 months, but achievement of a response appeared to have little effect on overall survival. In a randomized trial conducted by Eastern Cooperative Oncology Group (ECOG) and Southwest Oncology Group (SWOG), patients with MDS were treated with either low-dose cytarabine or supportive care. Patients in supportive care who progressed could cross over and receive treatment with cytarabine. There was no significant difference between the cytarabine and supportive-care groups with respect to overall survival, time to progression, or frequency of transformation to AML (Table 123.6).²³⁹

The hypomethylating agent 5-azacytidine (AzaC) has produced significant benefit (Table 123.6).^240,241 Since genes with methylated cytosines are poorly or not transcribed, a series of experiments by Christman and colleagues and Creusot and colleagues led to the development of a biochemical model that provided an explanation for the action of azacitidine as an inducer of differentiation through its effects on DNA methylation.^242,243 aza-C, once incorporated into DNA, covalently binds to DNA methyltransferase, the enzyme in mammalian cells responsible for methylation of newly synthesized DNA. This binding results in hypomethylated DNA distal to the binding point and leads to transcription of previously methylated quiescent genes. In patients with ß-thalassemia,²⁴⁴ treatment with AzaC resulted in an increase in fetal hemoglobin production, which was associated with hypomethylation in the region of the gamma globin chain gene. On the basis of this model, a trial testing the efficacy of AzaC in MDS was undertaken by the Cancer and Leukemia Group B (CALGB).²⁴⁵ Aza-C was administered at 75 mg/m²/d continuous intravenous infusion for 7 days repeated in 28-day cycles. Of 43 evaluable patients, 21 (49%) achieved a response; 12% had complete normalization of bone marrow and peripheral blood counts (CR); 25% had a ≥ 50% reduction in the initial percentage of BM blasts and a ≥ 50% restitution of the deficit from normal of all three peripheral blood counts (PR); and 12% had a ≥ 50% reduction in the deficit of ≥ 1 peripheral blood counts or a 50% or more reduction in transfusion requirements (Improved) (Table 123.8). The 37% trilineage response rate (CR+PR = ≥ 50% restitution of the deficit from normal of all three peripheral blood counts) is higher than for other therapies. Therapy-associated mortality occurred in 2% of patients. The overall median survival for all patients was 56 weeks, and the median time to relapse for patients achieving at least a PR was 14.9 months. The treatment was well tolerated, with grade 1 and 2 nausea being the most common side effect. The mechanism of action has yet to be defined. Only 33% of the patients required a dose reduction because of treatment induced myelosuppression, suggesting that the drug acted in the remainder, in part at least, through the induction of differentiation. In a second study in the CALGB, AzaC was administered at the same dose and schedule in an ambulatory regimen, with comparable efficacy and toxicity profiles.²⁴⁶ This led to the conduct of a phase III trial of AzaC, compared with supportive care (SC) in the CALGB. In a cross-over design, patients with progression could cross over to treatment after a minimum of four months in the SC group. Responses occurred in 63% on the AzaC arm (6% CR, 10% PR, 47% Improved), compared with 7% (Improved) in SC (p < .0001). The median time to leukemic transformation or death was 22 months for those on AzaC versus 12 months for SC (p = .0034). Probability of transformation to AML as the first event was lower on AzaC (11%) than SC (31%) (p = .003). Quality-of-life (QOL) assessment demonstrated significant major advantages in physical function, symptoms, and psychological state for patients initially on AzaC and for those on SC after cross-over to AzaC. Median survival for AzaC and SC (analyzed by intent-to-treat regardless of cross-over) was 18 and 14 months, respectively (p = .1). The probability of survival at 24 months was 41% for AzaC, compared with 25% for SC (p = .03), respectively. Results demonstrate that patients treated with AzaC had significantly higher response rates, improved QOL, delayed time to progression, improved survival at 24 months, delayed time to leukemic transformation or death and significantly reduced risk of transformation to AML, compared with supportive care.^240,241 Thus, AzaC is the only agent other than allogeneic bone marrow transplantation to alter the natural history of MDS. Furthermore, AzaC is not age restricted, as marrow transplantation is. Finally, the same rigorous response criteria that we developed for the first AzaC trial noted above, in the absence of uniform criteria, were utilized in the subsequent AzaC studies, including the randomized controlled trial (see Table 123.8).^240,245,246 The low response rate in the supportive care arm demonstrates that the criteria are sufficiently discriminating to filter out normal variation in count while specific enough to detect biologically important changes. Thus, the results of these studies validate the biologic utility of these criteria, which are the first to have a significant correlation with overall treatment outcome.

Azacitidine likely acts via a number of mechanisms. It can be a cytotoxic agent, but in vitro data indicate that it also functions as a biologic response modifier, affecting cytokine signal transduction pathways.^245,247 Another hypomethylating agent, decitabine, (2-deoxy-5-azacytidine) has also been evaluated in a smaller number of patients.²⁴⁸ Although response criteria are different, decitabine, like azacitidine, produces responses (see Table 123.7). However, there appear to be differences between the two agents, and the impact of decitabine on patient outcome remains to be determined with more extensive testing.

Retinoic acid and related compounds, highly effective inducers of differentiation in vitro,²⁴⁹ have been disappointing in their lack of efficacy in several clinical trials. In trials including some 235 patients reported in the literature, the overall response rate to cis-retinoic acid is 20%, and only one patient achieved a CR.^{191,250–252} Two randomized controlled trials have not demonstrated a difference in survival, compared with control, except in one small patient subgroup.^191,252 In that subgroup were patients with nonsideroblastic MDS and < 5% blasts in the bone marrow. They had a significantly prolonged survival, when compared with untreated controls. The survival of the control group was unusually short in comparison with what would normally be observed, however, and may well explain the reason for this isolated finding. All-trans-retinoic acid has been studied in 52 patients, with only 4 responses, and does not appear to have significant activity.²⁵³

Growth Factors

The hematopoietic growth factors are regulatory glycoproteins, which control the proliferation and differentiation of bone marrow stem cells.²⁵⁴ Several studies have defined the effects of GM-CSF in MDS. In a controlled study, patients were randomized to observation or treatment with rhGM-CSF.^255,256 Those treated had significant increases in neutrophils, eosinophils, monocytes, and lymphocytes, while the frequency of infections was decreased in comparison with those observed in the absence of this cytokine. There were no differences in platelet count, hemoglobin or transfusion requirements between the two groups. The risk of leukemic transformation appeared greatest for those patients with > 15% blasts in the bone marrow, and this may be a critical level with respect to leukemic transformation.²⁵⁷

Maintenance therapy, administering GM-CSF on a chronic basis, has met with mixed results, with some reports of benefit or even continued improvement in the granulocyte count, while other reports have shown a progressive deterioration in counts despite continued treatment.^258,259

Filgrastim (G-CSF) has also been evaluated. In a randomized controlled trial, 102 patients with RAEB or RAEB-T were treated with either G-CSF or supportive care.²⁶⁰ No differences in rates or time to progression to AML were seen between the two groups. There was no difference in survival for those with RAEB-T. However, among patients with RAEB, those in the treatment group had a significantly shortened survival, compared with patients in the control group, resulting in early termination of the study.

IL-3, another member of the family of hematopoietic growth factors, differs from G-CSF and GM-CSF, in so far as its stimulatory effects work on earlier hematopoietic stem cells, which have a multi-potential capacity for differentiation. Clinically, IL-3 appears similar to G- and GM-CSF in its effects on myeloid cells, has little impact on other lineages, and is significantly more toxic.^261,262

Human recombinant erythropoietin has also been studied in patients with MDS, with red cell responses being demonstrated in 20 to 25% of the patients tested.^106,263 Responses were confined to the erythroid lineage. A meta-analysis suggests that efficacy declines as bone marrow failure progresses.²⁶⁴ Those who have lower serum erythropoietin levels and less transfusion need are more likely to respond. The observation that in vitro erythropoiesis improved in patients treated with in vivo G-CSF led to two trials of erythropoietin and G-CSF in combination. The response rate ranged from 35 to 40% and appears to enhance the activity of erythropoietin. Serum erythropoietin levels in these trials, as in others, are of predictive value, with few responses in patients with levels above 500 U/L. In one other report, the response-enhancing effect of G-CSF was not substantiated.²⁶⁵ A randomized trial to further test this combination is currently underway.

For patients with a variant of MDS characterized by severe hypoplasia and pancytopenia, which may resemble severe aplastic anemia (hypoplastic MDS), administration of antithymocyte globulin produced responses in 11 of 25 patients treated (9 of 14 RA; 2 of 6 RAEB).²⁶⁶ Responses were characterized predominantly by loss of transfusion requirement, and 3 of 25 had normalization of their counts. There were no changes in the dysplastic features or in bone marrow cellularity. The effect may be mediated, in part, through an immunosuppressive effect alleviating a T-cell suppression of hematopoietic progenitors.²⁶⁷ Patients who are younger, have more cytopenias, and express CD-59 appear more likely to respond.

Amifostine, a phosphorylated aminothiol acts to protect normal tissues from the adverse effects of ionizing radiation and certain chemotherapy agents. On the basis of its in vitro effects on hematopoietic progenitors, 18 patients with MDS were treated, with reported hematologic improvement in 83%.²⁶⁸ Assessed according to the criteria utilized by the CALGB, as noted above, 10 (55%) would have been considered responders, 33% monolineage, 17% bilineage, and 6% trilineage. In a follow-up phase II study, amifostine at doses of 200 or 400 mg/m²/d three times weekly for 3 weeks, followed by 2 weeks of rest, was administered to 117 patients.²⁶⁹ A preliminary report on 75 evaluable patients demonstrated responses in 27 (36%). Responses occurred in all three lineages; however, trilineage responses in individual patients were not described. Ten of 47 patients whose bone marrows were centrally reviewed had a ≥ 50% decrease in the percentage of blasts with treatment. This correlated with the hematologic responses. Further follow-up and assessment and ultimately a randomized trial will be required to determine the impact on patient outcome.