Transient Leukaemia (TL) occurs frequently in newborn infants with Down Syndrome. Because in most cases the disease disappears spontaneously, it has been difficult to accept it as leukaemia, and as a result it has been referred by a variety of names including Transient Myeloproliferative Disorder and Transient Abnormal Myelopoiesis, suggesting that it is not leukaemia. In this review I shall present evidence that the disease is leukaemia and that although in most cases the course of the disease is transient in others it is life threatening. Also, I shall explore the nature of the leukaemic cell, its association with trisomy 21 and its relationship to other forms of leukaemia frequently found in children with Down Syndrome.
The clinical picture of transient leukaemia
Classic TL with spontaneous resolution
The original description of transient leukaemia (TL) and other early reports described this disorder as one in which there were no overt signs or symptoms of the disease. Indeed in most cases that is true. Children with Down Syndrome are found to have an elevated white blood count and primitive cells (‘blasts’) in the peripheral blood. Usually the blood count has been performed either as a routine study or because of an illness unrelated to the abnormal haematology. Upon examination there may either be no signs of the disease or there may be some enlargement of the liver and spleen or a skin rash (see below) in a minority of such cases. Usually the patient remains well and the leukaemic cells (‘blasts’) gradually decrease in number and disappear completely within the first 3 months of life.
TL with life-threatening complications
It has been recently reported in a prospective series that 19% of TL cases are complicated by life-threatening complications (Al-Kasim et al, 2002). There are two major forms.
Liver disease. This manifests as progressive obstructive jaundice with terminal liver failure (Miyauchi et al, 1992). Histologically there is hepatic fibrosis with evidence of infiltration by megakaryoblasts (Becroft & Zwi, 1990; Ruchelli et al, 1991). Approximately 50% of the cases will be progressive and fatal (Miyauchi et al, 1992; Al-Kasim et al, 2002). Others patients appear to recover completely with no long-term sequelae. Low-dose chemotherapy may be an effective treatment of the disease (Al-Kasim et al, 2002).
It has been suggested that the hepatic fibrosis is caused by the release of platelet-derived growth factor (PDG-B), which has been found in the tissues of these patients together with the infiltrating leukaemic cells (Terui et al, 1990; Hattori et al, 2001).
Cardio-pulmonary disease. This manifests as hydrops-like symptoms with pericardial effusions and ascites (Hendricks et al, 1993; Zipursky et al, 1996; Al-Kasim et al, 2002). The cause of this disorder is unknown, although there is evidence of leukaemic cell infiltration of the cardiac muscle (Zipursky et al, 1996). The disease manifests as pulmonary oedema and/or pericardial effusion; the disease may be progressive and fatal. There is evidence that low-dose chemotherapy may be curative (Doyle et al, 1995; Al-Kasim et al, 2002). Pericardial effusions are seen frequently in TL as isolated findings with no evidence of the cardio-pulmonary syndrome. In addition, isolated ascites has been reported (Shiffer & Natarajan, 2001). Leukaemic cells are found in both effusions. In one case studied by the author, basophils were the predominant cell in the effusion (Zipursky et al, 1997).
TL as a cause of intrauterine death
We described a series of four newborns with hydrops fetalis due to TL (Zipursky et al, 1996). Two of the infants were stillborn, one died after birth and one recovered. These four cases were among 40 Down Syndrome children born in one hospital. The incidence in the group was 10%. Smrcek et al (2001) reported a study of 79 Down Syndrome fetuses examined in utero using two-dimensional echocardiography. They found 11 (14%) of the cases had fetal hydrops. An isolated pericardial effusion has also been observed in a fetus of 34 weeks gestation (Hirashima et al, 2000). At birth there was haematological evidence of ‘Transient Abnormal Megakaryocytopoiesis’.
Skin lesions are seen frequently in TL. There is a description of three cases of vesiculopustular eruptions in Down Syndrome-related neonatal TL (Nijhawan et al, 2001). The skin lesions were described as vesiculo-papular and on biopsy leukaemic cells were found in both in the lesion leukaemic infiltrate and fluid.
TL with progression to leukaemia
As noted below, approximately 20% of TL cases that recover completely in the first months of life will subsequently develop acute megakaryoblastic leukaemia (AMKL) in the first 4 years of life (Doyle & Zipursky, 1994). In some cases, however, there is not a complete remission of the leukaemia. We have described one case in which TL improved with therapy, but blasts did not disappear completely and mild thrombocytopenia persisted (Al-Kasim et al, 2002). By 6 months of age the blast numbers increased, then a cytogenic abnormality (in addition to trisomy 21) appeared in the leukaemic cells and it was evident that the TL had merged directly into AMKL.
The treatment of TL
As noted above, some cases of TL have life-threatening complications. There are no controlled trials of therapy; however, in a small series it was suggested that treatment with low-dose cytosine arabinoside (10 mg/m2, twice a day for 7 d) can be curative (Al-Kasim et al, 2002). The use of low-dose therapy was based on experience in the treatment of AMKL in Down Syndrome in which low-dose cytosine arabinoside was curative (Tchernia et al, 1996; Zipursky, 1996).
The incidence of TL
In a small series it was found that 10% of newborn infants with Down Syndrome had blasts in their blood (Doyle & Zipursky, 1994). As screening was done by microscopic examination of blood films, it is possible that some cases with relatively few blasts would have been missed. It is likely that the incidence may be higher because it does not include fetuses with disease who died or who recovered in utero. As noted above, 14% of Down Syndrome fetuses, studied by ultrasound, had TL (Smrcek et al, 2001). We reported that in one hospital there were four cases of hydrops fetalis among 40 Down Syndrome newborns (Zipursky et al, 1996). These two reports suggest that the incidence of TL in fetuses, most of which died in utero, is approximately 10%. As screening of live newborns with Down Syndrome revealed an additional 10% of cases with TL, it may be concluded that approximately one in five fetuses with Down Syndrome have TL; many die before birth, many are asymptomatic, and some have severe disease in the neonatal period.
Mosaicism and TL
There are many reports of TL occurring in phenotypically normal children who have trisomy 21 mosaicism (Doyle et al, 1995; Slayton et al, 2002). When leukaemia occurs in these children, all the leukaemic cells have trisomy 21; in these patients assessment of skin biopsy fibroblasts or of PHA-stimulated lymphocytes, usually reveals a small percentage of cells that are trisomy 21. The percentage of trisomy 21 cells varies greatly and may be so low that they are not found in the non-leukaemic cells; in these cases mosaicism is restricted to the haematopoietic system. Slayton et al (2002) described a child with trisomy 21 mosaicism and TL in which only cells of the haematopoietic system had trisomy 21. A skin fibroblast culture and PHA stimulation of lymphocytes revealed no trisomy 21 cells. In children with trisomy 21 mosaicism and TL, all the leukaemic cells contain trisomy 21. This provides evidence that TL arises only in cells that have a third chromosome 21.
Although it has been stated that TL in patients with phenotypically normal trisomy 21 mosaics did not progress to AMKL, we have reported a case of TL in a phenotypically normal trisomy 21 mosaic with progression to AMKL (Doyle et al, 1995). The corollary of these observations is that TL (and probably AMKL) with trisomy 21 in a phenotypically normal child is likely to be the same form of leukaemia found in children with Down Syndrome.
The incidence of trisomy 21 mosaicism in newborn infants is approximately 1/2000–1/3000 (Hsu & Perlis, 1984; Worton & Stern, 1994). As the prevalence of Down Syndrome in newborn infants is 1/700, it is likely that TL occurs as readily in infants mosaic for trisomy 21 as in Down Syndrome. There are however, no definitive data to confirm this speculation.
The haematology of transient leukaemia
Peripheral blood cytology
The classic presentation of TL is one in which the haematology is normal except for the presence of primitive cells (‘blasts’) in the peripheral blood. These ‘blasts’ vary in number, from a very small percentage of the total leucocyte count to greater than 200 × 109/l (Zipursky et al, 1997). The blasts vary in size from 15 to 20 microns in diameter. They contain a small amount of basophilic cytoplasm with some cytoplasmic budding. The nuclei show minimal condensation and contain one to three nucleoli. The blasts are considered to be megakaryoblasts although, as discussed below, features of other cell lineages may be found. This is discussed in the next section.
Neutrophil and haemoglobin levels are normal. Platelet counts are usually normal; however, both thrombocytopenia and thrombocytosis are present in some cases with counts as low as 10 × 109/l to greater than 1000 × 109/l (Zipursky et al, 1997). In addition, macrothrombocytes and megakaryocyte fragments are found in blood films (Zipursky et al, 1999). TL leukaemic cells express the thrombopoietin receptor (c-mpl). The plasma levels of thrombopoietin are low in TL (Bonno et al, 1998) and are inversely related to the number of blasts in the peripheral blood. Presumably the low levels of thrombopoietin result from its binding to blast c-mpl and subsequent removal (Bonno et al, 1998).
Dysplastic nucleated red cells have been observed in the peripheral blood (Bozner, 2002) and some of these erythroid precursors have an abnormal morphology (i.e. dyserythropoietic). Of considerable interest is the presence of basophils in the peripheral blood. In one reported case of TL, the number of basophils in the peripheral blood was as high as 56 × 109/l (Worth et al, 1999).
The number of blasts declines during the first three post-natal months along with abnormalities in platelet number and morphology. Temporary increases in blast numbers may be seen during this time; however, it is rare for blasts to be found after 3 months of age. A case illustrating these changes is described elsewhere (Zipursky et al, 1997).
The leukaemic cells have been found to be negative when stained for peroxidase, Sudan Black, PAS and chloracetate esterase, but positive for acid phosphatase and non-specific esterase (with variable inhibition by sodium fluoride) (Athale et al, 2001).
The ultrastructure of the cell displays many features of megakaryoblasts (Eguchi et al, 1989; Zipursky et al, 1995). The cells may vary from primitive undifferentiated cells with no evidence of megakaryocyte differentiation to cells that contain demarcation membranes, alpha granules and megakaryocyte antigens, as demonstrated by immunogold labelling.
Surface antigen expression
There have been many studies of the antigens expressed on the surface of TL cells. All studies have shown that the megakaryocyte surface antigen, CD 41 [GpIIIa/GPIIb and CD 61 (GPIIb)] was detectable by immunolabelling followed by flow cytometry (Kurahashi et al, 1992; Yumura-Yagi et al, 1992; Girodon et al, 2000; Karandikar et al, 2001). Electron microscopic studies with immunogold labelling indicate that the most immature blasts may not express these antigens (Zipursky et al, 1995). The antigen CD34, a marker of a primitive haematopoietic precursor cell, is also expressed on the surface of TL blasts.
The leukaemic cells express the surface antigens CD7, which is a lymphoid marker, and CD 33, which is a myeloid marker. There is no evidence, in vivo, of lymphoid differentiation and, in the myeloid series, evidence only of basophilic and possibly eosinophilic differentiation. The expression of these antigens is also seen on primitive haematopoietic cells (Koike et al, 1987; Yumura-Yagi et al, 1992), which may explain their presence in the leukaemic cells of TL.
However, there is evidence that these cells express erythroid antigens. As discussed below, markers of erythroid differentiation (γ-globin and δ-aminolevulinate synthetase) have been observed by flow cytometry (Ito et al, 1995).
Bone marrow aspirate
Typically the bone marrow aspirate is cellular. Blasts are present in the marrow and their numbers are similar to, and directly related to, the percentage of blasts in the peripheral blood (Zipursky et al, 1997). This is most unlike other forms of leukaemia, because when the peripheral blood contains a large numbers of blasts, the marrow is virtually replaced by leukaemic cells. The late erythroid precursors may be abnormal in appearance, i.e. dyserythropoiesis. Megakaryocytes may be reduced in number, although this is typical of marrow aspirates in newborns. A few atypical, dysplastic, megakaryocytes may be seen.
Bone marrow biopsy
A distinctive feature of the bone marrow in TL is found in the marrow biopsy. An example of this is shown in the marrow biopsy of a newborn infant with TL and thrombocytopenia (Fig 1). The marrow contains large numbers of dysplastic megakaryocytes, a condition identical to that of the myelodysplasia observed in the early stages of acute megakaryoblastic leukaemia of Down Syndrome patients (Zipursky et al, 1994). The presence of increased numbers of megakaryocytes, many of which are dysplastic, in the presence of thrombocytopenia (Fig 1) is indicative of abnormal platelet formation. Additional evidence of abnormal megakaryocytopoiesis is the presence of megathrombocytes and megakaryocyte fragments in the peripheral blood (Zipursky et al, 1999).
The megakaryoblasts of Transient Leukaemia
The above observations, especially the ultrastructure and surface antigen expression, indicate that the leukaemic cells of TL have the properties of a precursor cell of megakaryocyte lineage, i.e. a megakaryoblast. However, the TL leukaemic cells also have properties of other cell lineages. An interpretation of the nature of the leukaemic cell is presented in a later section of this article.
Chromosomal analysis of transient leukaemia
All the leukaemic cells in TL of Down Syndrome patients or in phenotypically normal children with trisomy 21 mosaicism are trisomic for chromosome 21. In the majority of cases there are no other chromosomal abnormalities. There are, however, several reports of clonal chromosomal abnormalities in the leukaemic cells of TL. These include pentasomy 21 (Rogers et al, 1983), additional chromosomes 12 and 14 (Kounami et al, 1997), deletion of a chromosome (Polski et al, 2002), der(X;15)(p10;q10) (Duflos-Delaplace et al, 1999), an extra C chromosome (Honda et al, 1964) and polyploidy with 57 chromosomes (Engel et al, 1964).
The chromosomal abnormalities and leukaemic cells disappear at the same time. However, in those cases in which AMKL develops in the first 4 years of life, the abnormality seen in TL does not reappear in the leukaemic cells of AMKL (which have usually developed other chromosomal abnormalities). There are, however, important exceptions to that observation. Three of the chromosomal abnormalities found in TL, as noted above (Honda et al, 1964; Duflos-Delaplace et al, 1999; Polski et al, 2002), were present in the AMKL that appeared later in life. These findings provide evidence that the AMKL had arisen in one or more cells of the original TL leukaemic clone.
There has been one report of an intrachromosomal rearrangement of chromosome 21 in TL leukaemic cells (Kempski et al, 1998). This may be of importance because the region of chromosome 21 involved was the same as that in which interstitial deletions were found in five cases of megakaryoblastic leukaemia in Down Syndrome patients (Kempski et al, 1997).
Other forms of congenital leukaemia
Congenital leukaemia encompasses a rare group of disorders. A review of the cases reported in the literature (Bresters et al, 2002) found 117 cases of which 64% were acute non-lymphocytic leukaemias (ANLL) and 21% were acute lymphoblastic leukaemias (ALL). There was one case of French–American–British Cooperative Group classification M7 (due to a 1:22 translocation) and one case of type M6/M7 without trisomy 21. There have been no cases similar to TL other than those in patients with Down Syndrome or trisomy 21 mosaics. It should be added, however, that there are case reports in which phenotypically normal children have TL with trisomy 21 in the leukaemic cells but not in skin fibroblasts or PHA-stimulated lymphocytes (Faed et al, 1990). Presumably these children had mosaicism for chromosome 21 but this was restricted to cells of the haematological system, as noted above.
As there have been no reports of a TL-like disease in the absence of trisomy 21, it would appear that TL occurs only in newborn infants with either Down Syndrome or trisomy 21 mosaicism.
There is a report of non-identical twin newborns who both had a monoblastic form of leukaemia at birth with complete replacement of bone marrow and histological evidence of leukaemic infiltration of placental tissue (Mora et al, 2000). The disease disappeared spontaneously during the first month of life. These two cases illustrate that non-Down Syndrome leukaemia in newborns can undergo spontaneous cure, i.e. spontaneous cure does not rule out the diagnosis of leukaemia, which is an important consideration in evaluating the nature of TL in Down Syndrome. In the review described above (Bresters et al, 2002), six out of 75 cases of neonatal ANLL underwent spontaneous cure.
Transient leukaemia and acute megakaryoblastic leukaemia of down syndrome
As noted above, most cases of TL have a benign course in which the disease undergoes spontaneous remission and the patient is ‘cured’. However in approximately 20% of the cases that recover, Acute Megakaryoblastic Leukaemia (AMKL) develops within the first 4 months of life (Homans et al, 1993; Doyle & Zipursky, 1994). Although the incidence of ALL is higher in Down Syndrome than in non-sufferers, there is no evidence that TL predisposes to the development of ALL or any other form of leukaemia in Down Syndrome, other than AMKL.
Other than TL, AMKL is the commonest form of leukaemia in Down Syndrome patients. It occurs during the first 4 years of life and the incidence of AMKL in Down Syndrome is estimated to be approximately 500 times greater than in non-sufferers (Zipursky et al, 1992a). Although it is a form of megakaryoblastic leukaemia the disease is different from the two other forms of AMKL, which occur in childhood. One form occurs in infancy and is associated with a 1:22 translocation with formation of a distinct fusion gene (Ma et al, 2001a). That disease runs a rapid malignant course and shows a poor response to chemotherapy. The other form of megakaryocytic leukaemia, found in childhood, also has a poor response to chemotherapy (Schaison et al, 1990; Lange, 2000) and usually does not have the characteristic prolonged myelodysplastic phase of Down Syndrome AMKL (Lu et al, 1993; Zipursky et al, 1994). It can be concluded that the AMKL of Down Syndrome is distinct from other forms of megakaryoblastic leukaemia of childhood.
The leukaemic cell in Down Syndrome AMKL is similar to the leukaemic cell of TL in terms of morphology, cytochemistry and antigenic expression (Yumura-Yagi et al, 1994; Karandikar et al, 2001). The ultrastructure of the leukaemic cell of TL is very similar to that in AMKL; however, the TL cell has more evidence of megakaryocyte differentiation (Eguchi et al, 1989; Zipursky et al, 1995). Chromosomal abnormalities are more frequent in AMKL than in TL. In AMKL most cases have a chromosomal abnormality (Athale et al, 2001; Dastugue et al, 2002), the commonest being trisomy 8 (Zipursky et al, 1994; Athale et al, 2001). The nature of the chromosomal abnormalities in AMKL were considered by Ma et al (2001b). They studied the bone marrow aspirates in AMKL patients and found that the cytogenetic abnormalities (other than trisomy 21) were present only in a small number of cells, suggesting that the malignant changes evolved from a trisomy 21 population. They suggested that the chromosomal abnormalities seen in AMKL were a ‘secondary or evolutionary change’.
Although the leukaemic cells of TL and AMKL are similar, the course of disease is very different. As noted above, most cases of TL are benign and self-limiting whereas AMKL is a potentially lethal, malignant disease with no spontaneous cure. However, with standard either ANLL therapy or low-dose chemotherapy, the cure rate of Down Syndrome AMKL is very high (Zipursky et al, 1996; Lange, 2000). The benign and malignant character of TL and AMKL are reflected in studies of cell telomerase and P53 respectively.
Telomerase is an enzyme that maintains telomeres and is found in the majority of leukaemia cases in both children and adults. In one study, telomerase was found in over 50% of cases of AMKL whereas it was rare in the leukaemic cells of TL (Holt et al, 2002). Similarly, P53 mutations that are found in the majority of cases of adult leukaemia were not found in the leukaemic cells of TL but were present in two cases of AMKL (Malkin et al, 2000).
In most cases of TL the disease is a benign leukaemia; AMKL is a malignant leukaemia in which the course of the disease is potentially lethal, and the cells have features of malignancy: the expression of telomerase and mutations of p53. The available evidence suggests that TL and AMKL occur only in trisomy 21 cells. Furthermore, it is likely that AMKL occurs only in children who have had TL (see below). These observations suggest that TL and AMKL are different stages of a unique disease, which is distinct from any other form of leukaemia, perhaps best described as ‘Trisomy 21 leukaemia’.
Evidence that amkl arises from the leukaemic cells of transient leukaemia
The leukaemic cell of AMKL is similar to that of TL in morphology, cytochemistry, surface antigens and ultrastructure. Both cells show features of megakaryoblasts; however, both have erythroid and basophilic features, as noted below.
In both TL and AMKL, dysplastic megakaryocytopoiesis is a feature of the disease.
The incidence of TL and of AMKL in Down Syndrome patients is consistent with the theory that all cases of AMKL arise in children who have had TL. Thus the incidence of TL in Down Syndrome live births is approximately 10%. Twenty per cent of those children will develop AMKL, meaning that 1/50 children with Down Syndrome will develop AMKL. The incidence of AMKL in Down Syndrome is estimated to be 1/100–1/200 cases (Zipursky et al, 1992b). These calculations are crude estimates; however, they are consistent with the hypothesis that all cases of AMKL arise in children who presented with TL at birth.
There are reports of three children in whom the same chromosomal abnormalities were found in TL and in the leukaemic cells of AMKL that later developed. Honda et al (1964) described a patient in whom the leukaemic cells of TL and AMKL both contained an extra C chromosome. Further evidence that AMKL is derived from a clonal evolution of TL is furnished by the study of Polski et al (2002). They describe a case of TL, in a phenotypically normal child, that regressed spontaneously. However, at 3 months (at a time that the blast count had dropped to 3%) unstimulated peripheral blood showed trisomy 21and a new abnormality, an interstitial deletion of chromosome 13 (del (13)(q13q31). By 8 months age, there were no unstimulated dividing cells and Fluorescence in situ hybridization (FISH) analysis showed no evidence of either trisomy 21 or the interstitial deletion of 13. At 20 months, AMKL developed with a return of the trisomy 21; the leukaemic cells contained the interstitial deletion of chromosome 13 and a new abnormality of chromosome 17. Duflos-Delaplace et al (1999) described a case of TL in which the leukaemic cells had a chromosomal abnormality (der (X; 15)(p10;q10), which at 19 months of age, re-appeared in the leukaemic cells of AMKL.
It is likely that AMKL arises in some cells of the TL leukaemic cell population by ‘additional hits’; this clone then continues to proliferate while the other TL leukaemic cells disappear. The clone expands during the early months and years of life finally appearing as AMKL.
Evidence that transient leukaemia is leukaemia
It has been difficult to accept TL as leukaemia because it disappears spontaneously. However, as noted above in the section on ‘Congenital Leukaemia’, other forms of leukaemia can disappear spontaneously in the newborn period. Therefore, the principle that a spontaneously cured leukaemia cannot be leukaemia is not correct and, therefore, is not a reason to reject TL as a leukaemia. There is, in addition, the following direct evidence to support the conclusion that TL is leukaemia:
- 1The abnormal cell in TL is classified as a megakaryoblast with properties similar to the megakaryoblasts of the lethal AMKL that occurs in early childhood in Down Syndrome. Both TL and AMKL cells have features of megakaryocytic, erythroid and basophil lineages (see below).
- 2The disease may be lethal (as described above).
- 3At autopsy (Zipursky et al, 1996), in placental (de Tar et al, 2000) or skin biopsies (Nijhawan et al, 2001) there is evidence of leukaemic infiltration of tissues.
- 4As noted earlier there are several reports of chromosomal abnormalities in TL cases; this is consistent with a clonal (i.e. leukaemic) proliferation.
- 5Analysis of methylation patterns of the genes on the X-chromosome revealed that the leukaemic cells of TL are monoclonal, indicating proliferation from a single cell, i.e. a clonal proliferation consistent with a leukaemic process (Kurahashi et al, 1991; Miyashita et al, 1991).
Bresters et al (2002) defined Congenital Leukaemia as a leukaemia that fulfils the following diagnostic criteria:
a) Presentation in the first four weeks of life;
b) Proliferation of immature myeloid, lymphoid or erythroid cells;
c) Infiltration of these cells into non-haematopoietic tissues;
d) Absence of other diseases that may explain the proliferation
It is clear that TL fulfils all these criteria; in view of the above evidence we have concluded that TL is leukaemia (Zipursky & Doyle, 1993).
The nature of the leukaemic cells in transient leukaemia
- 1The morphology of the cell is consistent with the features of megakaryoblasts in other forms of megakaryoblastic leukaemia.
- 2The ultrastructure of the cell is consistent with that of a megakaryoblast with various degrees of differentiation.
- 3Identification of platelet peroxidase activity (Breton-Gorius & Guichard, 1972) by electron microscopy, in the megakaryoblasts of TL (Suda et al, 1988; Eguchi et al, 1989), established that the cells were of megakaryocyte lineage. As noted above, platelet antigens are demonstrable on the surface of the cell as evidenced by flow cytometry.
- 4The leukaemic cells express genes for the lineage-restricted GATA-1 that is expressed in megakaryocytic/erythroid precursor cells (Ito et al, 1995). GATA-1 is not expressed in lymphoid or other myeloid lineages.
As mentioned above, the bone marrow and peripheral blood in TL often contain abnormal erythroid precursors (dyserythropoiesis). Erythroid antigens have been detected in TL blasts; these include glycophorin (Bozner, 2002), as well as γ-globin and δ-aminolevulinate synthetase (Ito et al, 1995). A case of TL with predominant erythroid cells has been reported (Bozner, 2002). Electron microscopy-based identification of ferritin particles in the TL leukaemic cells is additional evidence of erythroid lineage involvement (Eguchi et al, 1989).
A case of transient leukaemia has been described in which there were massive numbers of basophils in the peripheral blood (Worth et al, 1999). In that case, basophil numbers peaked on the second day of life at 56 × 109/l. It should be noted that in the same case granulocytes with eosinophil-staining granules were also present in high numbers but were not considered to be eosinophils. Suda et al (1987) also described a case in which there were abnormally high numbers of basophils and eosinophils in the peripheral blood.
As noted earlier, we have studied one case in which the cells of a TL pericardial effusion were predominately basophils. In vitro culture of TL leukaemic cells produced basophils (Suda et al, 1987). Electron microscopy of TL cells revealed blasts which contained typical basophil granules (Eguchi et al, 1989; Zipursky et al, 1995).
The available evidence suggests that the leukaemic cell of TL is a precursor cell with the potential for producing cells of the megakaryocytic, erythroid and basophil series in vivo. It is not clear whether eosinophils are also produced in vivo. However, in vitro the leukaemic cells have not only the potential of producing cells of the megakaryocytic, erythroid and basophils series, but also cells of the neutrophil, monocyte and eosinophil series. It should be added that the production of basophils is particularly striking. Suda et al (1987) reported the growth of single cells and showed that the leukaemic cells could produce basophils, neutrophils, eosinophil, macrophages and erythroid elements. They also reported that approximately 50% of the cells grown in a culture of TL leukaemic cells were basophils, as identified by toluidine blue staining and electron microscopy.
The fact that a precursor cell with potential for megakaryocyte, erythroid and basophil production is part of normal haematopoiesis is discussed in a following section.
Culture of transient leukaemia cells
There are several reports of standard semisolid cultures of TL leukaemic cells. Most of these reports indicate that bone marrow cells grow to produce normal colonies (Rogers et al, 1983; de Alarcon et al, 1987). Liang et al (1993) described increased granulocyte–macrophage colony-forming units in culture. However, there is no evidence that in vitro growth characteristics are of any prognostic value, either in terms of the immediate outcome of the TL or the subsequent appearance of AMKL.
It can be concluded from the culture studies that the leukaemic cells can reproduce themselves, that the normal elements of the marrow will grow in vitro and that, in vitro, the blasts can differentiate into basophils, eosinophils, neutrophils and erythroid precursors.
Haematopoiesis and the megakaryoblastic leukaemias of down syndrome
There is evidence that in the normal process of haematopoiesis there is a precursor cell with the potential for forming both erythroid and megakaryocytic cells (Debili et al, 1996). The leukaemic cells of TL and AMKL have features of, and the potential for, in vitro formation of both these two cell lines. The evidence for this is provided above.
It is noted, however, that the leukaemic cells of TL and AMKL also have the potential of forming in vivo and in vitro cells of the basophilic series.
The evidence for a precursor cell of basophil/erythroid/megakaryocyte potential is not clear. What is clear, however, is that several transcription genes, GATA-1, GATA-2 and NF-E, are specific and necessary for megakaryocytopoiesis, erythropoiesis and basophil (or mast cell) production (Zon et al, 1993; Shivdasani, 2001).
Recently, a mutation in the gene for the transcription factor GATA-1 has been found in 6/6 patients with AMKL of Down Syndrome, suggesting therefore that this mutation may constitute a step in the pathogenesis of AMKL (Wechsler et al, 2002). GATA-1 is a transcription factor which is necessary for normal megakaryocytopoiesis (Shivdasani, 2001). It has been postulated that it is a negative regulator of megakaryocyte precursor proliferation, thereby allowing differentiation (Shivdasani, 2001). In the absence of GATA-1 (in GATA-1 knockout mice) there is an increase in megakaryocyte proliferation and both a delay in and abnormalities of megakaryocyte formation, with an ‘abnormally small and immature cytoplasm that harbours few platelet granules and highly disorganized internal membranes’ (Shivdasani, 2001). It is therefore of considerable interest that this also describes megakaryocytopoiesis in AMKL and is consistent with a mutation of GATA-1. The abnormal megakaryocytopoiesis in TL is similar to that in GATA-1-deficient mice and in AMKL. Accordingly, it remains to be determined whether mutations of GATA-1 are also present in TL. If so, the GATA-1 mutation would be related to the abnormal megakaryocytopoiesis of TL and AMKL and not to a later change in a TL cell as it mutates to the malignant cell of AMKL. Alternatively, it may be that the GATA-1 is not mutated in TL; this would suggest that the mutations found in AMKL were associated with the development of malignancy.
It can therefore be concluded that the benign and malignant leukaemic cells of TL and AMKL, respectively, are leukaemic proliferations of precursor cells with the potential for erythroid, megakaryocyte and basophil differentiation. The cells express transcription factor genes essential for the production of these three cell lines. The transition from the benign leukaemia of TL to the malignant form of this disease may be associated with changes in the genes responsible for the normal formation of these cell lines.
Transient Leukaemia and acute megakaryoblastic leukaemia are two stages of a unique form of leukaemia that occurs only in cells that are trisomic for chromosome 21. This disease does not occur in any other cell type. TL is distinct from other forms of congenital leukaemia and AMKL is distinct from other forms of megakaryoblastic leukaemia in childhood. These two disorders are different phases of a leukaemia that differs from all other forms of leukaemia and occurs only in cells that are constitutionally trisomy 21.
Note added in proof
Two recent studies found GATA-1 mutations in the cells of patients with transient leukaemia (Hitzler, J.K., Cheung, J., Yue Li, Y., Scherer, S.W. & Zipursky, A. (2003) Blood First Edition Paper, prepublished online Feb 13, 2003; DOI 10.1182/blood-2003-01-0013 Abstract; Mundschau, G., Gurbuxani, S., Gamis, A.S., Greene, M.E., Arceci, R.J. & Crispino, J.D. (2003) Blood First Edition Paper, prepublished online Jan 30, 2003; DOI 10.1182/blood-2002-12-3904 Abstract.