IN DECEMBER 1990, at the 22nd Annual Meeting of the American Society of Hematology, investigators from the Pediatric Oncology Group (POG; now part of Children’s Oncology Group [COG]) initially presented the unique responsiveness of acute myeloid leukemia (AML) in children with Down syndrome (DS) compared with AML in children without DS, followed by a more complete report in 1992.1,2 Publications before these two reports either commented on the rather dismal outcome of AML in children with DS3 or the lack of any significant difference in survival from children without DS,4 with one anecdotal exception—a comment published in 1989 by Lie,5 noting a possible superior outcome. Since then, the high curability of AML in children with DS has been the subject of numerous publications, supported by the molecular, biochemical, and pharmacologic bases for the markedly superior outcome compared with AML in children without DS. Although all 12 DS children with AML enrolled on the original study POG 8498 achieved complete remission and were long survivors, subsequent publications, while noting the continuing evidence for the high curability (>= 75% event-free survival [EFS] for DS-AML), raised concern for increased mortality with intensification of therapy. This high responsiveness to chemotherapy of AML in those with DS is especially remarkable, given that phenotypically it has the features of megakaryocytic leukemia (AMkL) and is often preceded by a history of myelodysplastic syndrome, both considered poor prognostic predictors of AML in individuals without DS.6,7
In this issue of the Journal of Clinical Oncology, Gamis et al8 examine the experience with AML in children with DS treated on CCG (Children’s Cancer Group, now also part of the COG) study 2891 and report that among children with DS with AML, age older than 2 years may be a relative poor-risk feature. Multivariate analysis of data showed that children with DS who were 2 years or older at diagnosis had an increased risk of relapse (odds ratio 4.9; P = .006). An implication is that in DS children over 2 years AML may be biologically similar to de novo AML, hence the poorer response. Several questions emerge as a result.
Is the Leukemia in DS Biologically Different Based on Age at Diagnosis?
A closer examination of the data of Gamis et al suggests that the risk varies within age groups: "Patients who had AML that was diagnosed between ages 0 to 2 years (n = 94) had a 6-year EFS of 86%, those older than 2 to 4 years (n = 58) had a 70% EFS, and those older than 4 years (n = 9) had a 28% EFS." Thus, for children with DS who are older than 4 years, the EFS is markedly inferior, and is the same as for non-DS children without DS with AML receiving the "standard timing 4 day course" of the dexamethasone, cytarabine, 6-thioguanine, etoposide, and rubidomycin (DCTER) regimen (EFS 21% standard induction- see abstract of Gamis et al8). No data are presented separately for the 2 to 3 and 3 to 4 years age groups, nor are there any details of the individual cases of those older than 4 years. Regardless, the markedly inferior outcome in those 4 years or older would create a "Will Rogers" phenomenon, whether the age cut is at 2 or 3 years, because of stage or risk migration.9 Since the overwhelming majority of children with DS with AML are younger than 3 years of age and only 14% (23 of 161) were older than 36 months in the Gamis et al report, one could argue that if age-related biologic differences in the nature of AML in children with DS existed, a better age cut may be beyond 36 or 48 months of age.
The increased risk for AML in children with DS is well known.6,10,11 Children with DS are also known to have a form of spontaneously resolving leukemia, which is variously referred to as transient myeloproliferative disorder (TMD) or transient leukemia. Zipursky et al10,11 have noted that infants with DS who had TMD have a higher risk for the later development of AMkL, usually by 3 years of age. Recent studies have shown that nonactivating mutations of GATA1 are present in virtually all cases of DS-AML (all of the studied cases to date were in those younger than 4 years) and that the same mutations are seen in TMD cases as well.12–16 Further, in paired samples of TMD and AMkL in the same child, identical GATA1 mutations were noted.13 Backtracking studies in childhood acute lymphocytic leukemia and AML suggest an in utero origin of a leukemic clone in the majority of cases, and a suggestion that the leukemic load at birth may be lower in those with leukemia diagnosed later in childhood.17–19 Thus, an important question to be pursued from the report of Gamis et al8 is whether DS-AML cases older than 2 years lack prior history of TMD and GATA1 mutations? If so, it would clearly indicate a different biologic origin and possibly a difference in response.
What Is the Optimal Therapy for DS-AML?
The superior outcome in DS became obvious only after the inclusion of high-dose cytarabine (Ara-C; HiDAC) in the treatment of childhood AML. Thus, the prevailing opinion is that two or more courses of HiDAC post–remission induction may be necessary for optimal therapy. However, the 100% EFS reported by the POG investigators, based on the POG 8498 AML regimen, has not been duplicated in subsequent studies by POG or by other groups.20–23 Successor studies to POG 8498 eliminated the historic combination course of prednisone, vincristine, methotrexate, and 6-mercaptopurine (POMP), plus the final four courses of conventional dose Ara-C, and sought to intensify the postinduction courses. At the same time, recognition of the unique responsiveness of AML in children with DS lead both to greater enrollment of DS cases on therapeutic studies and the recognition of risk for toxicity and the potential for mortality from infections.24,25 Gamis et al observed no increase in mortality rate in children with DS, but did observe high pulmonary toxicity during induction, including the need for ventilatory support and an increased incidence of mucositis and skin toxicity (perhaps from Ara-C) during intensification. Prolonged induction toxicity may result in premature patient withdrawal from a clinical trial because of concerns from both parents and physicians of putting a developmentally challenged child through excessive toxicity. Studies from Nordic countries and from Germany have shown that undertreated children with DS with AML had markedly inferior outcomes.21,22 Such vagaries in small cohorts may have a profound effect on the eventual outcomes, and it would therefore be of interest to know how many patients completed all courses of chemotherapy in the older age groups of children with DS in the CCG 2891 study. Results with intensively-timed DCTER arm of CCG 2891 in children with DS were worse than with standard-timed DCTER.6 Furthermore, Kojima et al26 obtained an 80% 8-year EFS with two to eight courses of conventional-dose daunorubicin or etoposide, and Ara-C in a 3+3+7-day combination. Hence, moderate-intensity therapy (as in the Japanese study or standard 3+7 combination of daunorubicin + Ara-C, followed by two courses of HiDAC for six doses, each given after full recovery of counts) seems to be the most optimal therapy.
Could the Difference in Outcome by Age Observed by Gamis et al Have a Pharmacologic or Biochemical Basis?
No studies were done, and hence, only speculations are possible. There are several genes on chromosome 21 that not only alter the sensitivity to Ara-C, but also to anthracyclines and other drugs in both specific and nonspecific manners.24,27–29
De novo modulation of cytarabine metabolism.
Taub et al29 noted that the expression levels of two chromosome-21 localized genes, cystathionine B-synthase (CBS), and superoxide dismutase (SOD1), were several fold higher than expected from gene dosage effect. Increased activity of CBS (21q22.1) has been linked to the low levels of homocysteine, methionine, s-adenosyl methionine, and a relative folate deficiency in DS. The relative folate deficiency related to CBS potentially primes the cells for Ara-C cytotoxicity. The net effect of increased CBS activity is a decreased generation of deoxythymidine triphosphate (dTTP). Low dTTP results in the release of the feedback inhibition of deoxycytidine deaminase, resulting in decreased levels of deoxycytidine monophosphate and deoxycytidine triphosphate (dCTP), while deoxycytidine kinase activity is upregulated because of low deoxycytidine monophosphate. In this setting, administration of Ara-C would result in a net increase in the generation of Ara-C triphosphate (Ara-CTP), with less competition from dCTP for binding to DNA and RNA polymerase.27 In studies by the Wayne State group, 27 DS-AML cells have low endogenous dCTP and are significantly more sensitive to Ara-C in the in vitro tetrazolium based cytotoxicity assays; there was a positive correlation of the Ara-C triphosphate values with the CBS transcript levels, which were increased up to 10-fold, and the IC50 values, as expected, had an inverse relationship to the CBS expression level.29
Increased multidrug sensitivity in DS.
In addition to unique endogenous modulation of Ara-C metabolism, DS-AML cells are also unusually sensitive to daunorubicin and other drugs.28,30–33 Chromosome-21 localized carbonyl reductase is involved in the metabolism of anthracyclines, converting the parent drug to a less potent but longer-lasting alcohol derivative (daunorubicinol). The DS cells may be primed to undergo drug-induced apoptosis, presumably due to the well-known increased generation of oxygen radicals in DS cells.24,34,35 The neuronal damage in DS has been linked to increased oxygen radical generation. The precise mechanism of the spontaneous increased oxygen radical generation in DS cells is unknown.36 Paradoxically, a modestly increased activity of superoxide dismutase 1, in the absence of concomitantly increased catalase, may actually be harmful presumably by generating increased amounts of the highly reactive hydroxyl radicals.20,35
Epigenetic influences on treatment response in DS-AML.
If future studies of DS-AML confirm the observations of Gamis et al, a closer look at the epigenetic phenomena that effect the cancer drug metabolism and the overall response to cytotoxic therapy as well, would have to be taken into account. For example, older children with DS are known to have neuroendocrine abnormalities and immune defects.37,38 Zinc deficiency, a well-known feature in older children with DS, has long been suspected as the culprit. There are more than 300 enzymes and 2,000 transcription factors (the list includes GATA1) that are zinc dependent.39 Of particular interest in the DS setting is the zinc dependency of SOD1 and several enzymes involved in B12-folate/thymidine synthetic pathway, including pyridoxal kinase (pyridoxal phosphatase is a cofactor for CBS), thymidine kinase, and 5' nucleotidase. The potential role of SOD1 and CBS has been discussed above. The defective immunity may contribute to the increased morbidity following intensive chemotherapy.
In summary, DS is a unique model for studying the association of leukemia (cancer) biology and therapeutic response. On the one hand, constitutional trisomy of chromosome 21 increases the risk for leukemia several-fold, but on the other, the leukemia is highly sensitive to chemotherapy. There is likely an Achilles’ heel associated with each of the somatic mutations and chromosomal aberrations, resulting in an increased risk for cancer; however, it is up to us to discover it. It is important to recognize that the molecular oncogenic event and the Achilles’ heel may not be one and the same.
From : http://jco.ascopubs.org/cgi/content/full/21/18/3385
.
Lefthit